Menu
▾
▴
xmlvm-developers — Used to discuss development of XMLVM itself. Not for developers using XMLVM.
You can subscribe to this list here.
| 2011 |
Jan
|
Feb
|
Mar
(38) |
Apr
(34) |
May
(20) |
Jun
(46) |
Jul
(6) |
Aug
(13) |
Sep
(50) |
Oct
(27) |
Nov
(10) |
Dec
(3) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 2012 |
Jan
(7) |
Feb
(3) |
Mar
(4) |
Apr
|
May
|
Jun
(2) |
Jul
(4) |
Aug
(3) |
Sep
(1) |
Oct
(2) |
Nov
|
Dec
|
| 2013 |
Jan
(1) |
Feb
|
Mar
(4) |
Apr
|
May
(1) |
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
| 2015 |
Jan
|
Feb
(1) |
Mar
|
Apr
|
May
|
Jun
|
Jul
|
Aug
|
Sep
|
Oct
|
Nov
|
Dec
|
|
From: Panayotis K. <pan...@pa...> - 2011-06-15 17:56:41
|
On Jun 15, 2011, at 5:33 PM, Arno Puder wrote: > >> Just wondering, since the user could always run everything through Proguard, >> before they even get to you. I was using Proguard for Android a year before >> they actually supported it. > > Probably a useful way to reduce the size. I don't think we should try to > replicate Proguards functionality in XMLVM. > > Arno It is very easy to reduce the size of a bunch of classes with proguard. I know how to do this on class level, but I 've read that you can also do it even on method level . |
|
From: Arno P. <ar...@pu...> - 2011-06-15 14:48:06
|
OK, I just removed the deprecated methods in NSObject. This permitted to do a *MUCH* more efficient implementation in the C backend via NSSelector. Arno On 6/6/11 2:50 PM, Panayotis Katsaloulis wrote: > The new NSSelector API is migrated to the Android backend too. > I've tested it and there seems to be no problem. > I didn't remove the deprecated methods of NSObject yet, just in case something is missing or is not properly working (although I doubt). > ------------------------------------------------------------------------------ > Simplify data backup and recovery for your virtual environment with vRanger. > Installation's a snap, and flexible recovery options mean your data is safe, > secure and there when you need it. Discover what all the cheering's about. > Get your free trial download today. > http://p.sf.net/sfu/quest-dev2dev2 > _______________________________________________ > Xmlvm-developers mailing list > Xml...@li... > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers |
|
From: Arno P. <ar...@pu...> - 2011-06-15 14:33:36
|
On 6/11/11 10:16 PM, Jeff Palmer wrote: > Saw this thread& all the system level responses. Android's use of Proguard > is at application level. Staying at the application level, I have a > question. If a xmlvm user had a large library jar, not RJE, that was only > slightly being used, would any of the un-used methods, end up in the final > program? I care as much about output size as build time. The only optimization XMLVM does is on class-level, i.e., once a class is pulled in to the compilation process, all its methods and fields will be included. XMLVM's dependency analysis is conservative: a class gets included the moment it gets referenced from the application. > Just wondering, since the user could always run everything through Proguard, > before they even get to you. I was using Proguard for Android a year before > they actually supported it. Probably a useful way to reduce the size. I don't think we should try to replicate Proguards functionality in XMLVM. Arno |
|
From: Arno P. <ar...@pu...> - 2011-06-14 12:47:41
|
On 6/14/11 5:00 AM, Panayotis Katsaloulis wrote: > I would like to ask a couple of things with C functions and XMLVM > > > 1) I want to create an XMLVM for some Foundation functions, like for example the following: > NSSearchPathForDirectoriesInDomains > NSTemporaryDirectory > In general, I am asking about some functions that do not really belong to a group but have distinct names. > What do you think is the best naming convention? > > For example, one would be something like > Foundation.functionName(...) This seems to be a good compromise. > 2) Are there any plans to properly support the "native" keyword with XMLVM? What do you mean with 'properly'? For the C backend you can call "ant gen-c-native-skeletons" which will generate default implementations of native methods in xmlvm/src/xmlvm2c/lib/native. The ObjC backend does not have this feature. > 3) Let's say that I have a library written in ObjC. What is the preferred way to "connect" this library with the C backend? > What if this library is written in C? No problems at all. ObjC is a superset of C. Just use the .m suffix for the filename and you can freely mix ObjC and C. Note that we already do this extensively for the Cocoa Touch wrappers (xmlvm/src/xmlvm2c/compat-lib/iphone). Arno |
|
From: Panayotis K. <pan...@pa...> - 2011-06-14 12:00:53
|
I would like to ask a couple of things with C functions and XMLVM 1) I want to create an XMLVM for some Foundation functions, like for example the following: NSSearchPathForDirectoriesInDomains NSTemporaryDirectory In general, I am asking about some functions that do not really belong to a group but have distinct names. What do you think is the best naming convention? For example, one would be something like Foundation.functionName(...) or NS.functionName(...) or NSTemporaryDirectory.get(...) or anything similar. What do you think? 2) Are there any plans to properly support the "native" keyword with XMLVM? 3) Let's say that I have a library written in ObjC. What is the preferred way to "connect" this library with the C backend? What if this library is written in C? |
|
From: Paul P. <bay...@gm...> - 2011-06-11 23:46:10
|
I have committed the change to sort the vtable entries. Paul On Sat, Jun 11, 2011 at 12:58 AM, Arno Puder <ar...@pu...> wrote: > > > On Jun 10, 2011, at 10:01 AM, Paul Poley <bay...@gm...> wrote: > > You can use rsync easily via Unix or on a Mac & with a little extra work in > Windows via Cygwin. > > > Hopefully we can integrate a similar behavior into the XMLVM pipeline. > > > The 2 obstacles: > > 1. The vtable entries are randomly ordered, so the files change > unnecessarily causing recompile. I just changed the > VtableOutputProcess.java to order alphabetically & the problem solved. > > > Can you please commit this to the HEAD? This would also be useful to make > gen-c-wrapper deterministic. > > > 1. Adding/removing String constants causes the index number to change > in other files, causing the other files to recompile unnecessarily. Would > that be easy to lock down index numbers in API classes? I.e. give them the > first numbers? > > > Good question. I had forgotten about the constant pool. Perhaps a similar > approach to what Markus suggested for the itables. I'll give it some > thought. > > Arno > > > > |
|
From: Jeff P. <jcp...@ro...> - 2011-06-11 20:16:53
|
Saw this thread & all the system level responses. Android's use of Proguard is at application level. Staying at the application level, I have a question. If a xmlvm user had a large library jar, not RJE, that was only slightly being used, would any of the un-used methods, end up in the final program? I care as much about output size as build time. Just wondering, since the user could always run everything through Proguard, before they even get to you. I was using Proguard for Android a year before they actually supported it. On 6/10/11 10:41 AM, "xml...@li..." <xml...@li...> wrote: > Date: Fri, 10 Jun 2011 12:00:50 +0300 > From: Panayotis Katsaloulis <pan...@pa...> > Subject: [xmlvm-dev] C backend and library > To: xml...@li... > Message-ID: <829...@pa...> > Content-Type: text/plain; charset=us-ascii > > Hello! > > I remember this discussion in the past, but I would like to bring it up again. >> From my understanding, the C backend right now computates on the fly the >> required classes and adds only those to the Xcode project. > Since now in C the handling of library should be more optimized, I think that > it would be very interesting to use the library (which will also reduce > compile time to minimum) and will make this class optimization obsolete. > > What do you think? > > |
|
From: Arno P. <ar...@pu...> - 2011-06-11 15:27:18
|
On Jun 11, 2011, at 3:37 AM, Panayotis Katsaloulis <pan...@pa...> wrote: > And now some questions: > > 1) Let's take for example the files java_util_ArrayList.m/h . This class gets overridden in my code by another class. Will the contents of java_util_ArrayList.m/h change, depending on the overriding class? Possibly yes. The method may not have been in the vtable but by overriding it, it needs to be added to the vtable of ArrayList. > 2) Is it possible (since most of it is generated code), to write code (probably with some casting) which will silence these 999+ warnings? The clean solution is to generate better code. In the meantime just select the option "inhibit all warnings". Arno > |
|
From: Panayotis K. <pan...@pa...> - 2011-06-11 10:37:46
|
First, some results: 1) The optimization of not touching a file if it remains the same is working and you could check it for yourself, in the SVN version. 2) C backend (with iFireworks) works fine in the iOS 5 3) Xcode still freezes, when the number of errors are over 1000. And now some questions: 1) Let's take for example the files java_util_ArrayList.m/h . This class gets overridden in my code by another class. Will the contents of java_util_ArrayList.m/h change, depending on the overriding class? 2) Is it possible (since most of it is generated code), to write code (probably with some casting) which will silence these 999+ warnings? |
|
From: Panayotis K. <pan...@pa...> - 2011-06-11 09:35:48
|
On Jun 10, 2011, at 8:01 PM, Paul Poley wrote: > I have already been using optimizations regarding this topic. I use "rsync" with options "-zvrc". There are 2 obstacles, one of which is very easy to overcome & the other I'll let someone else answer. > > This only speeds up XCode compile time, but that's the significant portion. The way it works is it XMLVM compiles as normal into a temporary directory. Then rsync copies all new or changed files to the destination directory. Any unchanged files are left there & so XCode does not recompile them. If you deleted any files, note they'll still be in the destination directory until you clean it out, but that doesn't matter most of the time. You could add an rsync option to delete them, but that would also delete the ".o" files & cause a full recompile. > > You can use rsync easily via Unix or on a Mac & with a little extra work in Windows via Cygwin. Sorry for the double post. This is more or less what is doing the build step I was talking about. |
|
From: Panayotis K. <pan...@pa...> - 2011-06-11 09:32:22
|
On Jun 10, 2011, at 7:26 PM, Sascha Haeberling wrote: > Ah gotcha. Any way to make that more stable so that not so many source files need to be changed? I am sure if this is relevant, but with the build patch I am proposing, there is a step which will touch a file ONLY if the contents of this file is changed. |
|
From: Arno P. <ar...@pu...> - 2011-06-11 05:58:26
|
On Jun 10, 2011, at 10:01 AM, Paul Poley <bay...@gm...> wrote: You can use rsync easily via Unix or on a Mac & with a little extra work in Windows via Cygwin. Hopefully we can integrate a similar behavior into the XMLVM pipeline. The 2 obstacles: 1. The vtable entries are randomly ordered, so the files change unnecessarily causing recompile. I just changed the VtableOutputProcess.java to order alphabetically & the problem solved. Can you please commit this to the HEAD? This would also be useful to make gen-c-wrapper deterministic. 1. Adding/removing String constants causes the index number to change in other files, causing the other files to recompile unnecessarily. Would that be easy to lock down index numbers in API classes? I.e. give them the first numbers? Good question. I had forgotten about the constant pool. Perhaps a similar approach to what Markus suggested for the itables. I'll give it some thought. Arno |
|
From: Paul P. <bay...@gm...> - 2011-06-10 17:02:01
|
I have already been using optimizations regarding this topic. I use "rsync" with options "-zvrc". There are 2 obstacles, one of which is very easy to overcome & the other I'll let someone else answer. This only speeds up XCode compile time, but that's the significant portion. The way it works is it XMLVM compiles as normal into a temporary directory. Then rsync copies all new or changed files to the destination directory. Any unchanged files are left there & so XCode does not recompile them. If you deleted any files, note they'll still be in the destination directory until you clean it out, but that doesn't matter most of the time. You could add an rsync option to delete them, but that would also delete the ".o" files & cause a full recompile. You can use rsync easily via Unix or on a Mac & with a little extra work in Windows via Cygwin. The 2 obstacles: 1. The vtable entries are randomly ordered, so the files change unnecessarily causing recompile. I just changed the VtableOutputProcess.java to order alphabetically & problem solved. 2. Adding/removing String constants causes the index number to change in other files, causing the other files to recompile unnecessarily. Would that be easy to lock down index numbers in API classes? I.e. give them the first numbers? Paul 2011/6/10 Sascha Haeberling <sa...@gm...> > I have to check when I am home, but I already do that in some cases. E.g. I > also ended up doing this for the JS output, because Qooxdoo had exactly the > same issue. It recompiled everything because of the timestamp changes. Once > that was fixed, it was sped up quite a bit > > // Sascha > > > On Fri, Jun 10, 2011 at 6:40 PM, Arno Puder <ar...@pu...> wrote: > >> >> Interfaces.h will change whenever you make any change to a Java interface >> or the way it is used. In this case the only option is a complete >> recompilation. >> >> What might already help if files that are written by an OutputProcess >> first check if the file already exists and has identical content. Of course >> that will slow down things a little more but might save time with Xcode. >> >> Arno >> >> >> On Jun 10, 2011, at 10:26 AM, Sascha Haeberling <sa...@gm...> >> wrote: >> >> Ah gotcha. Any way to make that more stable so that not so many source >> files need to be changed? >> >> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >> >>> >>> Yep. Xcode starts from scratch every time. But as you said, that is not >>> really Xcodes fault since we regenerate all the files every single time. >>> Note overwriting files is one thing, but remember that interfaces.h is >>> recomputed every single time which then requires all .m source files to be >>> recompiled as well. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 10:18 AM, Sascha Haeberling < <sa...@gm...> >>> sa...@gm...> wrote: >>> >>> So XCode starts totally from scratch every time you make a change? I >>> almost cannot believe that. Maybe it would help if we wouldn't change the >>> timestamps of files that haven't changed? >>> >>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >>> >>>> >>>> Real numbers would be helpful. The problem with caching is that we can >>>> only do this in XMLVM but not in Xcode. Xcode already take a long time to >>>> compile. How much time it takes to link against a huge library by comparison >>>> we don't know. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 10:03 AM, Sascha Haeberling < <sa...@gm...><sa...@gm...> >>>> sa...@gm...> wrote: >>>> >>>> I would really be interesting it getting some actual numbers. At work I >>>> often see huge binaries link forever. So if we put the whole JDK >>>> cross-compiled into one library, I wonder how long that would take. >>>> >>>> Also keep in mind that another alternative to speeding things up is to >>>> have proper caching of things at various stages. That could speed things up >>>> quite a bit I suspect. >>>> >>>> // Sascha >>>> >>>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...> >>>> ar...@pu...> >>>> >>>>> >>>>> I'm thinking of a "development mode" where no optimizations are >>>>> performed. For vtable this means that all methods go into the vtable by >>>>> default. That should be much easier than disabling itable optimizations. >>>>> >>>>> With regards to time savings: I suspect that a library would be >>>>> beneficial. Keep in mind that compile time includes XMLVM compile time as >>>>> well as Xcode compile time. XMLVM would only need to compile the application >>>>> without computing the transitive closure of all referenced classes (let >>>>> alone cross compiling these classes) and Xcode would "only" need to link a >>>>> huge library (where I suspect that that will be faster than compiling >>>>> everything). But of course I don't have empirical evidence. >>>>> >>>>> Arno >>>>> >>>>> >>>>> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...><mar...@gm...><mar...@gm...> >>>>> mar...@gm...> wrote: >>>>> >>>>> Hi, >>>>> >>>>> For the vtable optimization for non-overridden methods I don't see a >>>>> possibility to do this at all. If we compile everything into a library we >>>>> never know when a class of the application overrides something in the JDK, >>>>> etc. >>>>> >>>>> For the selector coloring we could implement a possibility to save the >>>>> end state of the colored graph for the library and use this information when >>>>> we color the selector graph of the application. >>>>> >>>>> But I agree with Sascha: it would be nice to know if this would speed >>>>> things up significantly before we start with something like that. >>>>> >>>>> Markus >>>>> >>>>> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...><sa...@gm...><sa...@gm...> >>>>> sa...@gm...> >>>>> >>>>>> Are we even sure that linking against such a huge library would really >>>>>> be that much faster? From my experience linking huge binaries can take a >>>>>> long time as well. >>>>>> >>>>>> In the end it looks like this: timeSaved = (timePipelineNow - >>>>>> timePipelineThen) - timeLinking >>>>>> >>>>>> If the times saved is small, the question is whether it makes sense >>>>>> investing time into it at all. Therefore it would be good to have some >>>>>> estimates first. >>>>>> >>>>>> // Sascha >>>>>> >>>>>> >>>>>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...><ar...@pu...> >>>>>> ar...@pu...> >>>>>> >>>>>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>>>>> library (.a) that speeds things compilation time. However, the >>>>>>> problem >>>>>>> is that (at least) for now it wouldn't work with the C backend. We do >>>>>>> optimizations that go beyond the simple decision "should this file be >>>>>>> included or not" (what the linker typically does, and which, btw, is >>>>>>> also performed by XMLVM at compile-time). We also optimize things >>>>>>> like >>>>>>> vtables, e.g., if a method never gets overridden, it will not end up >>>>>>> in the vtable. This knowledge, however, only exists if we take the >>>>>>> application into account. If the application overrides a library >>>>>>> method, then the cross-compilation of the library also needs to >>>>>>> change. >>>>>>> >>>>>>> I agree that for a regular development cycle it would make sense to >>>>>>> have something like a library and only do full optimizations for the >>>>>>> final release. We would need a way to cross-compile without >>>>>>> optimizations. One big problem I see here are interface methods >>>>>>> (because of the selector coloring scheme that is used), but I better >>>>>>> let Markus respond to this since he implemented those optimizations. >>>>>>> >>>>>>> Arno >>>>>>> >>>>>>> >>>>>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>>>>> < <pan...@pa...> <pan...@pa...><pan...@pa...><pan...@pa...> >>>>>>> pan...@pa...> wrote: >>>>>>> >>>>>>> > >>>>>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>>>>> > >>>>>>> >> Panayotis, >>>>>>> >> >>>>>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>>>>> library"? The C backend does compute all dependencies so that the number of >>>>>>> classes that are compiled and outputted in the end is at a minimum. Only >>>>>>> required classes will be pulled in. Why would you want to replace this and >>>>>>> with what? >>>>>>> >> >>>>>>> >> // Sascha >>>>>>> > >>>>>>> > >>>>>>> > This is from an old discussion. >>>>>>> > >>>>>>> > The idea is to compile the whole JDK and iPhone compatibility >>>>>>> library (and probably other dependencies) into a C library (i.e. in a form >>>>>>> of .a file) and compile the project against this library. >>>>>>> > Thus the Xcode project will compile much much faster and even the >>>>>>> production of the Xcode project will be faster (and smaller) since no files >>>>>>> from the JDK and the compatibility library will be required any more. >>>>>>> > >>>>>>> > This idea was also in the earlier days of ObjC backend, but due to >>>>>>> the way ObjC handled dependencies and late binding, the gain was far less. >>>>>>> > Now with the strong typed C language, things should be much better. >>>>>>> > >>>>>>> > For example, if a function is not used at all, the C compiler >>>>>>> understands that and never embeds it into the output binary - we have >>>>>>> optimizations in the method-level, instead of the class-level. >>>>>>> > >>>>>>> > >>>>>>> > I don't know how virtual functions are handled and probably Arno >>>>>>> could help with this. >>>>>>> > >>>>>>> > >>>>>>> > PS: I remember that ProGuard is also able to perform this type of >>>>>>> optimizations and it is already used for this reason in Android projects. >>>>>>> Probably we should also do something similar? >>>>>>> > >>>>>>> ------------------------------------------------------------------------------ >>>>>>> > EditLive Enterprise is the world's most technically advanced >>>>>>> content >>>>>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>>>>> > Editing and ensure content is compliant with Accessibility >>>>>>> Checking. >>>>>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>>>> > _______________________________________________ >>>>>>> > Xmlvm-developers mailing list >>>>>>> > <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>>>> Xml...@li... >>>>>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>>>> >>>>>>> >>>>>>> ------------------------------------------------------------------------------ >>>>>>> EditLive Enterprise is the world's most technically advanced content >>>>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>>>> _______________________________________________ >>>>>>> Xmlvm-developers mailing list >>>>>>> <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>>>> Xml...@li... >>>>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>>>> >>>>>> >>>>>> >>>>>> >>>>>> ------------------------------------------------------------------------------ >>>>>> EditLive Enterprise is the world's most technically advanced content >>>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>>> _______________________________________________ >>>>>> Xmlvm-developers mailing list >>>>>> <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>>> Xml...@li... >>>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>>> >>>>>> >>>>> >>>> >>> >> > > > ------------------------------------------------------------------------------ > EditLive Enterprise is the world's most technically advanced content > authoring tool. Experience the power of Track Changes, Inline Image > Editing and ensure content is compliant with Accessibility Checking. > http://p.sf.net/sfu/ephox-dev2dev > _______________________________________________ > Xmlvm-developers mailing list > Xml...@li... > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers > > |
|
From: Sascha H. <sa...@gm...> - 2011-06-10 16:47:53
|
I have to check when I am home, but I already do that in some cases. E.g. I also ended up doing this for the JS output, because Qooxdoo had exactly the same issue. It recompiled everything because of the timestamp changes. Once that was fixed, it was sped up quite a bit // Sascha On Fri, Jun 10, 2011 at 6:40 PM, Arno Puder <ar...@pu...> wrote: > > Interfaces.h will change whenever you make any change to a Java interface > or the way it is used. In this case the only option is a complete > recompilation. > > What might already help if files that are written by an OutputProcess first > check if the file already exists and has identical content. Of course that > will slow down things a little more but might save time with Xcode. > > Arno > > > On Jun 10, 2011, at 10:26 AM, Sascha Haeberling <sa...@gm...> wrote: > > Ah gotcha. Any way to make that more stable so that not so many source > files need to be changed? > > 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> > >> >> Yep. Xcode starts from scratch every time. But as you said, that is not >> really Xcodes fault since we regenerate all the files every single time. >> Note overwriting files is one thing, but remember that interfaces.h is >> recomputed every single time which then requires all .m source files to be >> recompiled as well. >> >> Arno >> >> >> On Jun 10, 2011, at 10:18 AM, Sascha Haeberling < <sa...@gm...> >> sa...@gm...> wrote: >> >> So XCode starts totally from scratch every time you make a change? I >> almost cannot believe that. Maybe it would help if we wouldn't change the >> timestamps of files that haven't changed? >> >> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >> >>> >>> Real numbers would be helpful. The problem with caching is that we can >>> only do this in XMLVM but not in Xcode. Xcode already take a long time to >>> compile. How much time it takes to link against a huge library by comparison >>> we don't know. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 10:03 AM, Sascha Haeberling < <sa...@gm...><sa...@gm...> >>> sa...@gm...> wrote: >>> >>> I would really be interesting it getting some actual numbers. At work I >>> often see huge binaries link forever. So if we put the whole JDK >>> cross-compiled into one library, I wonder how long that would take. >>> >>> Also keep in mind that another alternative to speeding things up is to >>> have proper caching of things at various stages. That could speed things up >>> quite a bit I suspect. >>> >>> // Sascha >>> >>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...> >>> ar...@pu...> >>> >>>> >>>> I'm thinking of a "development mode" where no optimizations are >>>> performed. For vtable this means that all methods go into the vtable by >>>> default. That should be much easier than disabling itable optimizations. >>>> >>>> With regards to time savings: I suspect that a library would be >>>> beneficial. Keep in mind that compile time includes XMLVM compile time as >>>> well as Xcode compile time. XMLVM would only need to compile the application >>>> without computing the transitive closure of all referenced classes (let >>>> alone cross compiling these classes) and Xcode would "only" need to link a >>>> huge library (where I suspect that that will be faster than compiling >>>> everything). But of course I don't have empirical evidence. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...><mar...@gm...><mar...@gm...> >>>> mar...@gm...> wrote: >>>> >>>> Hi, >>>> >>>> For the vtable optimization for non-overridden methods I don't see a >>>> possibility to do this at all. If we compile everything into a library we >>>> never know when a class of the application overrides something in the JDK, >>>> etc. >>>> >>>> For the selector coloring we could implement a possibility to save the >>>> end state of the colored graph for the library and use this information when >>>> we color the selector graph of the application. >>>> >>>> But I agree with Sascha: it would be nice to know if this would speed >>>> things up significantly before we start with something like that. >>>> >>>> Markus >>>> >>>> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...><sa...@gm...><sa...@gm...> >>>> sa...@gm...> >>>> >>>>> Are we even sure that linking against such a huge library would really >>>>> be that much faster? From my experience linking huge binaries can take a >>>>> long time as well. >>>>> >>>>> In the end it looks like this: timeSaved = (timePipelineNow - >>>>> timePipelineThen) - timeLinking >>>>> >>>>> If the times saved is small, the question is whether it makes sense >>>>> investing time into it at all. Therefore it would be good to have some >>>>> estimates first. >>>>> >>>>> // Sascha >>>>> >>>>> >>>>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...><ar...@pu...> >>>>> ar...@pu...> >>>>> >>>>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>>>> library (.a) that speeds things compilation time. However, the problem >>>>>> is that (at least) for now it wouldn't work with the C backend. We do >>>>>> optimizations that go beyond the simple decision "should this file be >>>>>> included or not" (what the linker typically does, and which, btw, is >>>>>> also performed by XMLVM at compile-time). We also optimize things like >>>>>> vtables, e.g., if a method never gets overridden, it will not end up >>>>>> in the vtable. This knowledge, however, only exists if we take the >>>>>> application into account. If the application overrides a library >>>>>> method, then the cross-compilation of the library also needs to >>>>>> change. >>>>>> >>>>>> I agree that for a regular development cycle it would make sense to >>>>>> have something like a library and only do full optimizations for the >>>>>> final release. We would need a way to cross-compile without >>>>>> optimizations. One big problem I see here are interface methods >>>>>> (because of the selector coloring scheme that is used), but I better >>>>>> let Markus respond to this since he implemented those optimizations. >>>>>> >>>>>> Arno >>>>>> >>>>>> >>>>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>>>> < <pan...@pa...> <pan...@pa...><pan...@pa...><pan...@pa...> >>>>>> pan...@pa...> wrote: >>>>>> >>>>>> > >>>>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>>>> > >>>>>> >> Panayotis, >>>>>> >> >>>>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>>>> library"? The C backend does compute all dependencies so that the number of >>>>>> classes that are compiled and outputted in the end is at a minimum. Only >>>>>> required classes will be pulled in. Why would you want to replace this and >>>>>> with what? >>>>>> >> >>>>>> >> // Sascha >>>>>> > >>>>>> > >>>>>> > This is from an old discussion. >>>>>> > >>>>>> > The idea is to compile the whole JDK and iPhone compatibility >>>>>> library (and probably other dependencies) into a C library (i.e. in a form >>>>>> of .a file) and compile the project against this library. >>>>>> > Thus the Xcode project will compile much much faster and even the >>>>>> production of the Xcode project will be faster (and smaller) since no files >>>>>> from the JDK and the compatibility library will be required any more. >>>>>> > >>>>>> > This idea was also in the earlier days of ObjC backend, but due to >>>>>> the way ObjC handled dependencies and late binding, the gain was far less. >>>>>> > Now with the strong typed C language, things should be much better. >>>>>> > >>>>>> > For example, if a function is not used at all, the C compiler >>>>>> understands that and never embeds it into the output binary - we have >>>>>> optimizations in the method-level, instead of the class-level. >>>>>> > >>>>>> > >>>>>> > I don't know how virtual functions are handled and probably Arno >>>>>> could help with this. >>>>>> > >>>>>> > >>>>>> > PS: I remember that ProGuard is also able to perform this type of >>>>>> optimizations and it is already used for this reason in Android projects. >>>>>> Probably we should also do something similar? >>>>>> > >>>>>> ------------------------------------------------------------------------------ >>>>>> > EditLive Enterprise is the world's most technically advanced content >>>>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>>>> > Editing and ensure content is compliant with Accessibility Checking. >>>>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>>> > _______________________________________________ >>>>>> > Xmlvm-developers mailing list >>>>>> > <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>>> Xml...@li... >>>>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>>> >>>>>> >>>>>> ------------------------------------------------------------------------------ >>>>>> EditLive Enterprise is the world's most technically advanced content >>>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>>> _______________________________________________ >>>>>> Xmlvm-developers mailing list >>>>>> <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>>> Xml...@li... >>>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>>> >>>>> >>>>> >>>>> >>>>> ------------------------------------------------------------------------------ >>>>> EditLive Enterprise is the world's most technically advanced content >>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>> _______________________________________________ >>>>> Xmlvm-developers mailing list >>>>> <Xml...@li...><Xml...@li...><Xml...@li...><Xml...@li...> >>>>> Xml...@li... >>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>> >>>>> >>>> >>> >> > |
|
From: Arno P. <ar...@pu...> - 2011-06-10 16:40:34
|
Interfaces.h will change whenever you make any change to a Java interface or the way it is used. In this case the only option is a complete recompilation. What might already help if files that are written by an OutputProcess first check if the file already exists and has identical content. Of course that will slow down things a little more but might save time with Xcode. Arno On Jun 10, 2011, at 10:26 AM, Sascha Haeberling <sa...@gm...> wrote: Ah gotcha. Any way to make that more stable so that not so many source files need to be changed? 2011/6/10 Arno Puder <ar...@pu...> > > Yep. Xcode starts from scratch every time. But as you said, that is not > really Xcodes fault since we regenerate all the files every single time. > Note overwriting files is one thing, but remember that interfaces.h is > recomputed every single time which then requires all .m source files to be > recompiled as well. > > Arno > > > On Jun 10, 2011, at 10:18 AM, Sascha Haeberling <sa...@gm...> wrote: > > So XCode starts totally from scratch every time you make a change? I almost > cannot believe that. Maybe it would help if we wouldn't change the > timestamps of files that haven't changed? > > 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> > >> >> Real numbers would be helpful. The problem with caching is that we can >> only do this in XMLVM but not in Xcode. Xcode already take a long time to >> compile. How much time it takes to link against a huge library by comparison >> we don't know. >> >> Arno >> >> >> On Jun 10, 2011, at 10:03 AM, Sascha Haeberling < <sa...@gm...> >> sa...@gm...> wrote: >> >> I would really be interesting it getting some actual numbers. At work I >> often see huge binaries link forever. So if we put the whole JDK >> cross-compiled into one library, I wonder how long that would take. >> >> Also keep in mind that another alternative to speeding things up is to >> have proper caching of things at various stages. That could speed things up >> quite a bit I suspect. >> >> // Sascha >> >> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >> >>> >>> I'm thinking of a "development mode" where no optimizations are >>> performed. For vtable this means that all methods go into the vtable by >>> default. That should be much easier than disabling itable optimizations. >>> >>> With regards to time savings: I suspect that a library would be >>> beneficial. Keep in mind that compile time includes XMLVM compile time as >>> well as Xcode compile time. XMLVM would only need to compile the application >>> without computing the transitive closure of all referenced classes (let >>> alone cross compiling these classes) and Xcode would "only" need to link a >>> huge library (where I suspect that that will be faster than compiling >>> everything). But of course I don't have empirical evidence. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...><mar...@gm...> >>> mar...@gm...> wrote: >>> >>> Hi, >>> >>> For the vtable optimization for non-overridden methods I don't see a >>> possibility to do this at all. If we compile everything into a library we >>> never know when a class of the application overrides something in the JDK, >>> etc. >>> >>> For the selector coloring we could implement a possibility to save the >>> end state of the colored graph for the library and use this information when >>> we color the selector graph of the application. >>> >>> But I agree with Sascha: it would be nice to know if this would speed >>> things up significantly before we start with something like that. >>> >>> Markus >>> >>> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...><sa...@gm...> >>> sa...@gm...> >>> >>>> Are we even sure that linking against such a huge library would really >>>> be that much faster? From my experience linking huge binaries can take a >>>> long time as well. >>>> >>>> In the end it looks like this: timeSaved = (timePipelineNow - >>>> timePipelineThen) - timeLinking >>>> >>>> If the times saved is small, the question is whether it makes sense >>>> investing time into it at all. Therefore it would be good to have some >>>> estimates first. >>>> >>>> // Sascha >>>> >>>> >>>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...> >>>> ar...@pu...> >>>> >>>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>>> library (.a) that speeds things compilation time. However, the problem >>>>> is that (at least) for now it wouldn't work with the C backend. We do >>>>> optimizations that go beyond the simple decision "should this file be >>>>> included or not" (what the linker typically does, and which, btw, is >>>>> also performed by XMLVM at compile-time). We also optimize things like >>>>> vtables, e.g., if a method never gets overridden, it will not end up >>>>> in the vtable. This knowledge, however, only exists if we take the >>>>> application into account. If the application overrides a library >>>>> method, then the cross-compilation of the library also needs to >>>>> change. >>>>> >>>>> I agree that for a regular development cycle it would make sense to >>>>> have something like a library and only do full optimizations for the >>>>> final release. We would need a way to cross-compile without >>>>> optimizations. One big problem I see here are interface methods >>>>> (because of the selector coloring scheme that is used), but I better >>>>> let Markus respond to this since he implemented those optimizations. >>>>> >>>>> Arno >>>>> >>>>> >>>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>>> < <pan...@pa...> <pan...@pa...><pan...@pa...> >>>>> pan...@pa...> wrote: >>>>> >>>>> > >>>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>>> > >>>>> >> Panayotis, >>>>> >> >>>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>>> library"? The C backend does compute all dependencies so that the number of >>>>> classes that are compiled and outputted in the end is at a minimum. Only >>>>> required classes will be pulled in. Why would you want to replace this and >>>>> with what? >>>>> >> >>>>> >> // Sascha >>>>> > >>>>> > >>>>> > This is from an old discussion. >>>>> > >>>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>>> file) and compile the project against this library. >>>>> > Thus the Xcode project will compile much much faster and even the >>>>> production of the Xcode project will be faster (and smaller) since no files >>>>> from the JDK and the compatibility library will be required any more. >>>>> > >>>>> > This idea was also in the earlier days of ObjC backend, but due to >>>>> the way ObjC handled dependencies and late binding, the gain was far less. >>>>> > Now with the strong typed C language, things should be much better. >>>>> > >>>>> > For example, if a function is not used at all, the C compiler >>>>> understands that and never embeds it into the output binary - we have >>>>> optimizations in the method-level, instead of the class-level. >>>>> > >>>>> > >>>>> > I don't know how virtual functions are handled and probably Arno >>>>> could help with this. >>>>> > >>>>> > >>>>> > PS: I remember that ProGuard is also able to perform this type of >>>>> optimizations and it is already used for this reason in Android projects. >>>>> Probably we should also do something similar? >>>>> > >>>>> ------------------------------------------------------------------------------ >>>>> > EditLive Enterprise is the world's most technically advanced content >>>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>>> > Editing and ensure content is compliant with Accessibility Checking. >>>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>> > _______________________________________________ >>>>> > Xmlvm-developers mailing list >>>>> > <Xml...@li...><Xml...@li...><Xml...@li...> >>>>> Xml...@li... >>>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>> >>>>> >>>>> ------------------------------------------------------------------------------ >>>>> EditLive Enterprise is the world's most technically advanced content >>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>> _______________________________________________ >>>>> Xmlvm-developers mailing list >>>>> <Xml...@li...><Xml...@li...><Xml...@li...> >>>>> Xml...@li... >>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>> >>>> >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...><Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>> >> > |
|
From: Sascha H. <sa...@gm...> - 2011-06-10 16:26:38
|
Ah gotcha. Any way to make that more stable so that not so many source files need to be changed? 2011/6/10 Arno Puder <ar...@pu...> > > Yep. Xcode starts from scratch every time. But as you said, that is not > really Xcodes fault since we regenerate all the files every single time. > Note overwriting files is one thing, but remember that interfaces.h is > recomputed every single time which then requires all .m source files to be > recompiled as well. > > Arno > > > On Jun 10, 2011, at 10:18 AM, Sascha Haeberling <sa...@gm...> wrote: > > So XCode starts totally from scratch every time you make a change? I almost > cannot believe that. Maybe it would help if we wouldn't change the > timestamps of files that haven't changed? > > 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> > >> >> Real numbers would be helpful. The problem with caching is that we can >> only do this in XMLVM but not in Xcode. Xcode already take a long time to >> compile. How much time it takes to link against a huge library by comparison >> we don't know. >> >> Arno >> >> >> On Jun 10, 2011, at 10:03 AM, Sascha Haeberling < <sa...@gm...> >> sa...@gm...> wrote: >> >> I would really be interesting it getting some actual numbers. At work I >> often see huge binaries link forever. So if we put the whole JDK >> cross-compiled into one library, I wonder how long that would take. >> >> Also keep in mind that another alternative to speeding things up is to >> have proper caching of things at various stages. That could speed things up >> quite a bit I suspect. >> >> // Sascha >> >> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >> >>> >>> I'm thinking of a "development mode" where no optimizations are >>> performed. For vtable this means that all methods go into the vtable by >>> default. That should be much easier than disabling itable optimizations. >>> >>> With regards to time savings: I suspect that a library would be >>> beneficial. Keep in mind that compile time includes XMLVM compile time as >>> well as Xcode compile time. XMLVM would only need to compile the application >>> without computing the transitive closure of all referenced classes (let >>> alone cross compiling these classes) and Xcode would "only" need to link a >>> huge library (where I suspect that that will be faster than compiling >>> everything). But of course I don't have empirical evidence. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...><mar...@gm...> >>> mar...@gm...> wrote: >>> >>> Hi, >>> >>> For the vtable optimization for non-overridden methods I don't see a >>> possibility to do this at all. If we compile everything into a library we >>> never know when a class of the application overrides something in the JDK, >>> etc. >>> >>> For the selector coloring we could implement a possibility to save the >>> end state of the colored graph for the library and use this information when >>> we color the selector graph of the application. >>> >>> But I agree with Sascha: it would be nice to know if this would speed >>> things up significantly before we start with something like that. >>> >>> Markus >>> >>> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...><sa...@gm...> >>> sa...@gm...> >>> >>>> Are we even sure that linking against such a huge library would really >>>> be that much faster? From my experience linking huge binaries can take a >>>> long time as well. >>>> >>>> In the end it looks like this: timeSaved = (timePipelineNow - >>>> timePipelineThen) - timeLinking >>>> >>>> If the times saved is small, the question is whether it makes sense >>>> investing time into it at all. Therefore it would be good to have some >>>> estimates first. >>>> >>>> // Sascha >>>> >>>> >>>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...><ar...@pu...> >>>> ar...@pu...> >>>> >>>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>>> library (.a) that speeds things compilation time. However, the problem >>>>> is that (at least) for now it wouldn't work with the C backend. We do >>>>> optimizations that go beyond the simple decision "should this file be >>>>> included or not" (what the linker typically does, and which, btw, is >>>>> also performed by XMLVM at compile-time). We also optimize things like >>>>> vtables, e.g., if a method never gets overridden, it will not end up >>>>> in the vtable. This knowledge, however, only exists if we take the >>>>> application into account. If the application overrides a library >>>>> method, then the cross-compilation of the library also needs to >>>>> change. >>>>> >>>>> I agree that for a regular development cycle it would make sense to >>>>> have something like a library and only do full optimizations for the >>>>> final release. We would need a way to cross-compile without >>>>> optimizations. One big problem I see here are interface methods >>>>> (because of the selector coloring scheme that is used), but I better >>>>> let Markus respond to this since he implemented those optimizations. >>>>> >>>>> Arno >>>>> >>>>> >>>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>>> < <pan...@pa...> <pan...@pa...><pan...@pa...> >>>>> pan...@pa...> wrote: >>>>> >>>>> > >>>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>>> > >>>>> >> Panayotis, >>>>> >> >>>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>>> library"? The C backend does compute all dependencies so that the number of >>>>> classes that are compiled and outputted in the end is at a minimum. Only >>>>> required classes will be pulled in. Why would you want to replace this and >>>>> with what? >>>>> >> >>>>> >> // Sascha >>>>> > >>>>> > >>>>> > This is from an old discussion. >>>>> > >>>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>>> file) and compile the project against this library. >>>>> > Thus the Xcode project will compile much much faster and even the >>>>> production of the Xcode project will be faster (and smaller) since no files >>>>> from the JDK and the compatibility library will be required any more. >>>>> > >>>>> > This idea was also in the earlier days of ObjC backend, but due to >>>>> the way ObjC handled dependencies and late binding, the gain was far less. >>>>> > Now with the strong typed C language, things should be much better. >>>>> > >>>>> > For example, if a function is not used at all, the C compiler >>>>> understands that and never embeds it into the output binary - we have >>>>> optimizations in the method-level, instead of the class-level. >>>>> > >>>>> > >>>>> > I don't know how virtual functions are handled and probably Arno >>>>> could help with this. >>>>> > >>>>> > >>>>> > PS: I remember that ProGuard is also able to perform this type of >>>>> optimizations and it is already used for this reason in Android projects. >>>>> Probably we should also do something similar? >>>>> > >>>>> ------------------------------------------------------------------------------ >>>>> > EditLive Enterprise is the world's most technically advanced content >>>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>>> > Editing and ensure content is compliant with Accessibility Checking. >>>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>> > _______________________________________________ >>>>> > Xmlvm-developers mailing list >>>>> > <Xml...@li...><Xml...@li...><Xml...@li...> >>>>> Xml...@li... >>>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>> >>>>> >>>>> ------------------------------------------------------------------------------ >>>>> EditLive Enterprise is the world's most technically advanced content >>>>> authoring tool. Experience the power of Track Changes, Inline Image >>>>> Editing and ensure content is compliant with Accessibility Checking. >>>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>>> http://p.sf.net/sfu/ephox-dev2dev >>>>> _______________________________________________ >>>>> Xmlvm-developers mailing list >>>>> <Xml...@li...><Xml...@li...><Xml...@li...> >>>>> Xml...@li... >>>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>>> >>>> >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...><Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>> >> > |
|
From: Arno P. <ar...@pu...> - 2011-06-10 16:23:14
|
Yep. Xcode starts from scratch every time. But as you said, that is not really Xcodes fault since we regenerate all the files every single time. Note overwriting files is one thing, but remember that interfaces.h is recomputed every single time which then requires all .m source files to be recompiled as well. Arno On Jun 10, 2011, at 10:18 AM, Sascha Haeberling <sa...@gm...> wrote: So XCode starts totally from scratch every time you make a change? I almost cannot believe that. Maybe it would help if we wouldn't change the timestamps of files that haven't changed? 2011/6/10 Arno Puder <ar...@pu...> > > Real numbers would be helpful. The problem with caching is that we can only > do this in XMLVM but not in Xcode. Xcode already take a long time to > compile. How much time it takes to link against a huge library by comparison > we don't know. > > Arno > > > On Jun 10, 2011, at 10:03 AM, Sascha Haeberling <sa...@gm...> wrote: > > I would really be interesting it getting some actual numbers. At work I > often see huge binaries link forever. So if we put the whole JDK > cross-compiled into one library, I wonder how long that would take. > > Also keep in mind that another alternative to speeding things up is to have > proper caching of things at various stages. That could speed things up quite > a bit I suspect. > > // Sascha > > 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> > >> >> I'm thinking of a "development mode" where no optimizations are performed. >> For vtable this means that all methods go into the vtable by default. That >> should be much easier than disabling itable optimizations. >> >> With regards to time savings: I suspect that a library would be >> beneficial. Keep in mind that compile time includes XMLVM compile time as >> well as Xcode compile time. XMLVM would only need to compile the application >> without computing the transitive closure of all referenced classes (let >> alone cross compiling these classes) and Xcode would "only" need to link a >> huge library (where I suspect that that will be faster than compiling >> everything). But of course I don't have empirical evidence. >> >> Arno >> >> >> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...> >> mar...@gm...> wrote: >> >> Hi, >> >> For the vtable optimization for non-overridden methods I don't see a >> possibility to do this at all. If we compile everything into a library we >> never know when a class of the application overrides something in the JDK, >> etc. >> >> For the selector coloring we could implement a possibility to save the end >> state of the colored graph for the library and use this information when we >> color the selector graph of the application. >> >> But I agree with Sascha: it would be nice to know if this would speed >> things up significantly before we start with something like that. >> >> Markus >> >> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...> >> sa...@gm...> >> >>> Are we even sure that linking against such a huge library would really be >>> that much faster? From my experience linking huge binaries can take a long >>> time as well. >>> >>> In the end it looks like this: timeSaved = (timePipelineNow - >>> timePipelineThen) - timeLinking >>> >>> If the times saved is small, the question is whether it makes sense >>> investing time into it at all. Therefore it would be good to have some >>> estimates first. >>> >>> // Sascha >>> >>> >>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >>> >>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>> library (.a) that speeds things compilation time. However, the problem >>>> is that (at least) for now it wouldn't work with the C backend. We do >>>> optimizations that go beyond the simple decision "should this file be >>>> included or not" (what the linker typically does, and which, btw, is >>>> also performed by XMLVM at compile-time). We also optimize things like >>>> vtables, e.g., if a method never gets overridden, it will not end up >>>> in the vtable. This knowledge, however, only exists if we take the >>>> application into account. If the application overrides a library >>>> method, then the cross-compilation of the library also needs to >>>> change. >>>> >>>> I agree that for a regular development cycle it would make sense to >>>> have something like a library and only do full optimizations for the >>>> final release. We would need a way to cross-compile without >>>> optimizations. One big problem I see here are interface methods >>>> (because of the selector coloring scheme that is used), but I better >>>> let Markus respond to this since he implemented those optimizations. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>> < <pan...@pa...> <pan...@pa...> >>>> pan...@pa...> wrote: >>>> >>>> > >>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>> > >>>> >> Panayotis, >>>> >> >>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>> library"? The C backend does compute all dependencies so that the number of >>>> classes that are compiled and outputted in the end is at a minimum. Only >>>> required classes will be pulled in. Why would you want to replace this and >>>> with what? >>>> >> >>>> >> // Sascha >>>> > >>>> > >>>> > This is from an old discussion. >>>> > >>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>> file) and compile the project against this library. >>>> > Thus the Xcode project will compile much much faster and even the >>>> production of the Xcode project will be faster (and smaller) since no files >>>> from the JDK and the compatibility library will be required any more. >>>> > >>>> > This idea was also in the earlier days of ObjC backend, but due to the >>>> way ObjC handled dependencies and late binding, the gain was far less. >>>> > Now with the strong typed C language, things should be much better. >>>> > >>>> > For example, if a function is not used at all, the C compiler >>>> understands that and never embeds it into the output binary - we have >>>> optimizations in the method-level, instead of the class-level. >>>> > >>>> > >>>> > I don't know how virtual functions are handled and probably Arno could >>>> help with this. >>>> > >>>> > >>>> > PS: I remember that ProGuard is also able to perform this type of >>>> optimizations and it is already used for this reason in Android projects. >>>> Probably we should also do something similar? >>>> > >>>> ------------------------------------------------------------------------------ >>>> > EditLive Enterprise is the world's most technically advanced content >>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>> > Editing and ensure content is compliant with Accessibility Checking. >>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> > _______________________________________________ >>>> > Xmlvm-developers mailing list >>>> > <Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev> <http://p.sf.net/sfu/ephox-dev2dev> >>> http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...><Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >> > |
|
From: Sascha H. <sa...@gm...> - 2011-06-10 16:18:45
|
So XCode starts totally from scratch every time you make a change? I almost cannot believe that. Maybe it would help if we wouldn't change the timestamps of files that haven't changed? 2011/6/10 Arno Puder <ar...@pu...> > > Real numbers would be helpful. The problem with caching is that we can only > do this in XMLVM but not in Xcode. Xcode already take a long time to > compile. How much time it takes to link against a huge library by comparison > we don't know. > > Arno > > > On Jun 10, 2011, at 10:03 AM, Sascha Haeberling <sa...@gm...> wrote: > > I would really be interesting it getting some actual numbers. At work I > often see huge binaries link forever. So if we put the whole JDK > cross-compiled into one library, I wonder how long that would take. > > Also keep in mind that another alternative to speeding things up is to have > proper caching of things at various stages. That could speed things up quite > a bit I suspect. > > // Sascha > > 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> > >> >> I'm thinking of a "development mode" where no optimizations are performed. >> For vtable this means that all methods go into the vtable by default. That >> should be much easier than disabling itable optimizations. >> >> With regards to time savings: I suspect that a library would be >> beneficial. Keep in mind that compile time includes XMLVM compile time as >> well as Xcode compile time. XMLVM would only need to compile the application >> without computing the transitive closure of all referenced classes (let >> alone cross compiling these classes) and Xcode would "only" need to link a >> huge library (where I suspect that that will be faster than compiling >> everything). But of course I don't have empirical evidence. >> >> Arno >> >> >> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <<mar...@gm...> >> mar...@gm...> wrote: >> >> Hi, >> >> For the vtable optimization for non-overridden methods I don't see a >> possibility to do this at all. If we compile everything into a library we >> never know when a class of the application overrides something in the JDK, >> etc. >> >> For the selector coloring we could implement a possibility to save the end >> state of the colored graph for the library and use this information when we >> color the selector graph of the application. >> >> But I agree with Sascha: it would be nice to know if this would speed >> things up significantly before we start with something like that. >> >> Markus >> >> 2011/6/10 Sascha Haeberling < <sa...@gm...> <sa...@gm...> >> sa...@gm...> >> >>> Are we even sure that linking against such a huge library would really be >>> that much faster? From my experience linking huge binaries can take a long >>> time as well. >>> >>> In the end it looks like this: timeSaved = (timePipelineNow - >>> timePipelineThen) - timeLinking >>> >>> If the times saved is small, the question is whether it makes sense >>> investing time into it at all. Therefore it would be good to have some >>> estimates first. >>> >>> // Sascha >>> >>> >>> 2011/6/10 Arno Puder < <ar...@pu...> <ar...@pu...>ar...@pu...> >>> >>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>> library (.a) that speeds things compilation time. However, the problem >>>> is that (at least) for now it wouldn't work with the C backend. We do >>>> optimizations that go beyond the simple decision "should this file be >>>> included or not" (what the linker typically does, and which, btw, is >>>> also performed by XMLVM at compile-time). We also optimize things like >>>> vtables, e.g., if a method never gets overridden, it will not end up >>>> in the vtable. This knowledge, however, only exists if we take the >>>> application into account. If the application overrides a library >>>> method, then the cross-compilation of the library also needs to >>>> change. >>>> >>>> I agree that for a regular development cycle it would make sense to >>>> have something like a library and only do full optimizations for the >>>> final release. We would need a way to cross-compile without >>>> optimizations. One big problem I see here are interface methods >>>> (because of the selector coloring scheme that is used), but I better >>>> let Markus respond to this since he implemented those optimizations. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>> < <pan...@pa...> <pan...@pa...> >>>> pan...@pa...> wrote: >>>> >>>> > >>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>> > >>>> >> Panayotis, >>>> >> >>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>> library"? The C backend does compute all dependencies so that the number of >>>> classes that are compiled and outputted in the end is at a minimum. Only >>>> required classes will be pulled in. Why would you want to replace this and >>>> with what? >>>> >> >>>> >> // Sascha >>>> > >>>> > >>>> > This is from an old discussion. >>>> > >>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>> file) and compile the project against this library. >>>> > Thus the Xcode project will compile much much faster and even the >>>> production of the Xcode project will be faster (and smaller) since no files >>>> from the JDK and the compatibility library will be required any more. >>>> > >>>> > This idea was also in the earlier days of ObjC backend, but due to the >>>> way ObjC handled dependencies and late binding, the gain was far less. >>>> > Now with the strong typed C language, things should be much better. >>>> > >>>> > For example, if a function is not used at all, the C compiler >>>> understands that and never embeds it into the output binary - we have >>>> optimizations in the method-level, instead of the class-level. >>>> > >>>> > >>>> > I don't know how virtual functions are handled and probably Arno could >>>> help with this. >>>> > >>>> > >>>> > PS: I remember that ProGuard is also able to perform this type of >>>> optimizations and it is already used for this reason in Android projects. >>>> Probably we should also do something similar? >>>> > >>>> ------------------------------------------------------------------------------ >>>> > EditLive Enterprise is the world's most technically advanced content >>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>> > Editing and ensure content is compliant with Accessibility Checking. >>>> > <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> > _______________________________________________ >>>> > Xmlvm-developers mailing list >>>> > <Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev><http://p.sf.net/sfu/ephox-dev2dev> >>>> http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...><Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev> <http://p.sf.net/sfu/ephox-dev2dev> >>> http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...><Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers><https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >> > |
|
From: Arno P. <ar...@pu...> - 2011-06-10 16:17:23
|
I think that is a good idea! For development mode the size of the itable does not matter. Arno On Jun 10, 2011, at 10:02 AM, Markus Neubrand <mar...@gm...> wrote: Forgot to mention: Alternatively we could color the app with a safe offset (e.g. 100) to avoid clashes in "development mode". But that would blow up tib sizes quite a bit. 2011/6/10 Markus Neubrand <mar...@gm...> > We would still need to store the selector graph: A class in the app could > implement its own interface as well as implement a JDK interface at the same > time. For this case I need to know how the original JDK interface was > colored to avoid clashes. > > For the vtable optimization implementing the "development mode" should be > simple. > > > 2011/6/10 Arno Puder <ar...@pu...> > >> >> I'm thinking of a "development mode" where no optimizations are performed. >> For vtable this means that all methods go into the vtable by default. That >> should be much easier than disabling itable optimizations. >> >> With regards to time savings: I suspect that a library would be >> beneficial. Keep in mind that compile time includes XMLVM compile time as >> well as Xcode compile time. XMLVM would only need to compile the application >> without computing the transitive closure of all referenced classes (let >> alone cross compiling these classes) and Xcode would "only" need to link a >> huge library (where I suspect that that will be faster than compiling >> everything). But of course I don't have empirical evidence. >> >> Arno >> >> >> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> >> wrote: >> >> Hi, >> >> For the vtable optimization for non-overridden methods I don't see a >> possibility to do this at all. If we compile everything into a library we >> never know when a class of the application overrides something in the JDK, >> etc. >> >> For the selector coloring we could implement a possibility to save the end >> state of the colored graph for the library and use this information when we >> color the selector graph of the application. >> >> But I agree with Sascha: it would be nice to know if this would speed >> things up significantly before we start with something like that. >> >> Markus >> >> 2011/6/10 Sascha Haeberling < <sa...@gm...>sa...@gm...> >> >>> Are we even sure that linking against such a huge library would really be >>> that much faster? From my experience linking huge binaries can take a long >>> time as well. >>> >>> In the end it looks like this: timeSaved = (timePipelineNow - >>> timePipelineThen) - timeLinking >>> >>> If the times saved is small, the question is whether it makes sense >>> investing time into it at all. Therefore it would be good to have some >>> estimates first. >>> >>> // Sascha >>> >>> >>> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >>> >>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>> library (.a) that speeds things compilation time. However, the problem >>>> is that (at least) for now it wouldn't work with the C backend. We do >>>> optimizations that go beyond the simple decision "should this file be >>>> included or not" (what the linker typically does, and which, btw, is >>>> also performed by XMLVM at compile-time). We also optimize things like >>>> vtables, e.g., if a method never gets overridden, it will not end up >>>> in the vtable. This knowledge, however, only exists if we take the >>>> application into account. If the application overrides a library >>>> method, then the cross-compilation of the library also needs to >>>> change. >>>> >>>> I agree that for a regular development cycle it would make sense to >>>> have something like a library and only do full optimizations for the >>>> final release. We would need a way to cross-compile without >>>> optimizations. One big problem I see here are interface methods >>>> (because of the selector coloring scheme that is used), but I better >>>> let Markus respond to this since he implemented those optimizations. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>> < <pan...@pa...>pan...@pa...> wrote: >>>> >>>> > >>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>> > >>>> >> Panayotis, >>>> >> >>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>> library"? The C backend does compute all dependencies so that the number of >>>> classes that are compiled and outputted in the end is at a minimum. Only >>>> required classes will be pulled in. Why would you want to replace this and >>>> with what? >>>> >> >>>> >> // Sascha >>>> > >>>> > >>>> > This is from an old discussion. >>>> > >>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>> file) and compile the project against this library. >>>> > Thus the Xcode project will compile much much faster and even the >>>> production of the Xcode project will be faster (and smaller) since no files >>>> from the JDK and the compatibility library will be required any more. >>>> > >>>> > This idea was also in the earlier days of ObjC backend, but due to the >>>> way ObjC handled dependencies and late binding, the gain was far less. >>>> > Now with the strong typed C language, things should be much better. >>>> > >>>> > For example, if a function is not used at all, the C compiler >>>> understands that and never embeds it into the output binary - we have >>>> optimizations in the method-level, instead of the class-level. >>>> > >>>> > >>>> > I don't know how virtual functions are handled and probably Arno could >>>> help with this. >>>> > >>>> > >>>> > PS: I remember that ProGuard is also able to perform this type of >>>> optimizations and it is already used for this reason in Android projects. >>>> Probably we should also do something similar? >>>> > >>>> ------------------------------------------------------------------------------ >>>> > EditLive Enterprise is the world's most technically advanced content >>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>> > Editing and ensure content is compliant with Accessibility Checking. >>>> > <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>>> > _______________________________________________ >>>> > Xmlvm-developers mailing list >>>> > <Xml...@li...> >>>> Xml...@li... >>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >> > |
|
From: Arno P. <ar...@pu...> - 2011-06-10 16:16:33
|
Real numbers would be helpful. The problem with caching is that we can only do this in XMLVM but not in Xcode. Xcode already take a long time to compile. How much time it takes to link against a huge library by comparison we don't know. Arno On Jun 10, 2011, at 10:03 AM, Sascha Haeberling <sa...@gm...> wrote: I would really be interesting it getting some actual numbers. At work I often see huge binaries link forever. So if we put the whole JDK cross-compiled into one library, I wonder how long that would take. Also keep in mind that another alternative to speeding things up is to have proper caching of things at various stages. That could speed things up quite a bit I suspect. // Sascha 2011/6/10 Arno Puder <ar...@pu...> > > I'm thinking of a "development mode" where no optimizations are performed. > For vtable this means that all methods go into the vtable by default. That > should be much easier than disabling itable optimizations. > > With regards to time savings: I suspect that a library would be beneficial. > Keep in mind that compile time includes XMLVM compile time as well as Xcode > compile time. XMLVM would only need to compile the application without > computing the transitive closure of all referenced classes (let alone cross > compiling these classes) and Xcode would "only" need to link a huge library > (where I suspect that that will be faster than compiling everything). But of > course I don't have empirical evidence. > > Arno > > > On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> > wrote: > > Hi, > > For the vtable optimization for non-overridden methods I don't see a > possibility to do this at all. If we compile everything into a library we > never know when a class of the application overrides something in the JDK, > etc. > > For the selector coloring we could implement a possibility to save the end > state of the colored graph for the library and use this information when we > color the selector graph of the application. > > But I agree with Sascha: it would be nice to know if this would speed > things up significantly before we start with something like that. > > Markus > > 2011/6/10 Sascha Haeberling < <sa...@gm...>sa...@gm...> > >> Are we even sure that linking against such a huge library would really be >> that much faster? From my experience linking huge binaries can take a long >> time as well. >> >> In the end it looks like this: timeSaved = (timePipelineNow - >> timePipelineThen) - timeLinking >> >> If the times saved is small, the question is whether it makes sense >> investing time into it at all. Therefore it would be good to have some >> estimates first. >> >> // Sascha >> >> >> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >> >>> Yes, we've had this discussion before. Basically creating a Unix-like >>> library (.a) that speeds things compilation time. However, the problem >>> is that (at least) for now it wouldn't work with the C backend. We do >>> optimizations that go beyond the simple decision "should this file be >>> included or not" (what the linker typically does, and which, btw, is >>> also performed by XMLVM at compile-time). We also optimize things like >>> vtables, e.g., if a method never gets overridden, it will not end up >>> in the vtable. This knowledge, however, only exists if we take the >>> application into account. If the application overrides a library >>> method, then the cross-compilation of the library also needs to >>> change. >>> >>> I agree that for a regular development cycle it would make sense to >>> have something like a library and only do full optimizations for the >>> final release. We would need a way to cross-compile without >>> optimizations. One big problem I see here are interface methods >>> (because of the selector coloring scheme that is used), but I better >>> let Markus respond to this since he implemented those optimizations. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>> < <pan...@pa...>pan...@pa...> wrote: >>> >>> > >>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>> > >>> >> Panayotis, >>> >> >>> >> sorry, I didn't understand what you mean. What do you mean by "the >>> library"? The C backend does compute all dependencies so that the number of >>> classes that are compiled and outputted in the end is at a minimum. Only >>> required classes will be pulled in. Why would you want to replace this and >>> with what? >>> >> >>> >> // Sascha >>> > >>> > >>> > This is from an old discussion. >>> > >>> > The idea is to compile the whole JDK and iPhone compatibility library >>> (and probably other dependencies) into a C library (i.e. in a form of .a >>> file) and compile the project against this library. >>> > Thus the Xcode project will compile much much faster and even the >>> production of the Xcode project will be faster (and smaller) since no files >>> from the JDK and the compatibility library will be required any more. >>> > >>> > This idea was also in the earlier days of ObjC backend, but due to the >>> way ObjC handled dependencies and late binding, the gain was far less. >>> > Now with the strong typed C language, things should be much better. >>> > >>> > For example, if a function is not used at all, the C compiler >>> understands that and never embeds it into the output binary - we have >>> optimizations in the method-level, instead of the class-level. >>> > >>> > >>> > I don't know how virtual functions are handled and probably Arno could >>> help with this. >>> > >>> > >>> > PS: I remember that ProGuard is also able to perform this type of >>> optimizations and it is already used for this reason in Android projects. >>> Probably we should also do something similar? >>> > >>> ------------------------------------------------------------------------------ >>> > EditLive Enterprise is the world's most technically advanced content >>> > authoring tool. Experience the power of Track Changes, Inline Image >>> > Editing and ensure content is compliant with Accessibility Checking. >>> > <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> > _______________________________________________ >>> > Xmlvm-developers mailing list >>> > <Xml...@li...> >>> Xml...@li... >>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >> >> >> >> ------------------------------------------------------------------------------ >> EditLive Enterprise is the world's most technically advanced content >> authoring tool. Experience the power of Track Changes, Inline Image >> Editing and ensure content is compliant with Accessibility Checking. >> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >> _______________________________________________ >> Xmlvm-developers mailing list >> <Xml...@li...> >> Xml...@li... >> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> >> > |
|
From: Sascha H. <sa...@gm...> - 2011-06-10 16:03:07
|
I would really be interesting it getting some actual numbers. At work I often see huge binaries link forever. So if we put the whole JDK cross-compiled into one library, I wonder how long that would take. Also keep in mind that another alternative to speeding things up is to have proper caching of things at various stages. That could speed things up quite a bit I suspect. // Sascha 2011/6/10 Arno Puder <ar...@pu...> > > I'm thinking of a "development mode" where no optimizations are performed. > For vtable this means that all methods go into the vtable by default. That > should be much easier than disabling itable optimizations. > > With regards to time savings: I suspect that a library would be beneficial. > Keep in mind that compile time includes XMLVM compile time as well as Xcode > compile time. XMLVM would only need to compile the application without > computing the transitive closure of all referenced classes (let alone cross > compiling these classes) and Xcode would "only" need to link a huge library > (where I suspect that that will be faster than compiling everything). But of > course I don't have empirical evidence. > > Arno > > > On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> > wrote: > > Hi, > > For the vtable optimization for non-overridden methods I don't see a > possibility to do this at all. If we compile everything into a library we > never know when a class of the application overrides something in the JDK, > etc. > > For the selector coloring we could implement a possibility to save the end > state of the colored graph for the library and use this information when we > color the selector graph of the application. > > But I agree with Sascha: it would be nice to know if this would speed > things up significantly before we start with something like that. > > Markus > > 2011/6/10 Sascha Haeberling < <sa...@gm...>sa...@gm...> > >> Are we even sure that linking against such a huge library would really be >> that much faster? From my experience linking huge binaries can take a long >> time as well. >> >> In the end it looks like this: timeSaved = (timePipelineNow - >> timePipelineThen) - timeLinking >> >> If the times saved is small, the question is whether it makes sense >> investing time into it at all. Therefore it would be good to have some >> estimates first. >> >> // Sascha >> >> >> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >> >>> Yes, we've had this discussion before. Basically creating a Unix-like >>> library (.a) that speeds things compilation time. However, the problem >>> is that (at least) for now it wouldn't work with the C backend. We do >>> optimizations that go beyond the simple decision "should this file be >>> included or not" (what the linker typically does, and which, btw, is >>> also performed by XMLVM at compile-time). We also optimize things like >>> vtables, e.g., if a method never gets overridden, it will not end up >>> in the vtable. This knowledge, however, only exists if we take the >>> application into account. If the application overrides a library >>> method, then the cross-compilation of the library also needs to >>> change. >>> >>> I agree that for a regular development cycle it would make sense to >>> have something like a library and only do full optimizations for the >>> final release. We would need a way to cross-compile without >>> optimizations. One big problem I see here are interface methods >>> (because of the selector coloring scheme that is used), but I better >>> let Markus respond to this since he implemented those optimizations. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>> < <pan...@pa...>pan...@pa...> wrote: >>> >>> > >>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>> > >>> >> Panayotis, >>> >> >>> >> sorry, I didn't understand what you mean. What do you mean by "the >>> library"? The C backend does compute all dependencies so that the number of >>> classes that are compiled and outputted in the end is at a minimum. Only >>> required classes will be pulled in. Why would you want to replace this and >>> with what? >>> >> >>> >> // Sascha >>> > >>> > >>> > This is from an old discussion. >>> > >>> > The idea is to compile the whole JDK and iPhone compatibility library >>> (and probably other dependencies) into a C library (i.e. in a form of .a >>> file) and compile the project against this library. >>> > Thus the Xcode project will compile much much faster and even the >>> production of the Xcode project will be faster (and smaller) since no files >>> from the JDK and the compatibility library will be required any more. >>> > >>> > This idea was also in the earlier days of ObjC backend, but due to the >>> way ObjC handled dependencies and late binding, the gain was far less. >>> > Now with the strong typed C language, things should be much better. >>> > >>> > For example, if a function is not used at all, the C compiler >>> understands that and never embeds it into the output binary - we have >>> optimizations in the method-level, instead of the class-level. >>> > >>> > >>> > I don't know how virtual functions are handled and probably Arno could >>> help with this. >>> > >>> > >>> > PS: I remember that ProGuard is also able to perform this type of >>> optimizations and it is already used for this reason in Android projects. >>> Probably we should also do something similar? >>> > >>> ------------------------------------------------------------------------------ >>> > EditLive Enterprise is the world's most technically advanced content >>> > authoring tool. Experience the power of Track Changes, Inline Image >>> > Editing and ensure content is compliant with Accessibility Checking. >>> > <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> > _______________________________________________ >>> > Xmlvm-developers mailing list >>> > <Xml...@li...> >>> Xml...@li... >>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >> >> >> >> ------------------------------------------------------------------------------ >> EditLive Enterprise is the world's most technically advanced content >> authoring tool. Experience the power of Track Changes, Inline Image >> Editing and ensure content is compliant with Accessibility Checking. >> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >> _______________________________________________ >> Xmlvm-developers mailing list >> <Xml...@li...> >> Xml...@li... >> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> >> > |
|
From: Markus N. <mar...@gm...> - 2011-06-10 16:02:38
|
Forgot to mention: Alternatively we could color the app with a safe offset (e.g. 100) to avoid clashes in "development mode". But that would blow up tib sizes quite a bit. 2011/6/10 Markus Neubrand <mar...@gm...> > We would still need to store the selector graph: A class in the app could > implement its own interface as well as implement a JDK interface at the same > time. For this case I need to know how the original JDK interface was > colored to avoid clashes. > > For the vtable optimization implementing the "development mode" should be > simple. > > > 2011/6/10 Arno Puder <ar...@pu...> > >> >> I'm thinking of a "development mode" where no optimizations are performed. >> For vtable this means that all methods go into the vtable by default. That >> should be much easier than disabling itable optimizations. >> >> With regards to time savings: I suspect that a library would be >> beneficial. Keep in mind that compile time includes XMLVM compile time as >> well as Xcode compile time. XMLVM would only need to compile the application >> without computing the transitive closure of all referenced classes (let >> alone cross compiling these classes) and Xcode would "only" need to link a >> huge library (where I suspect that that will be faster than compiling >> everything). But of course I don't have empirical evidence. >> >> Arno >> >> >> On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> >> wrote: >> >> Hi, >> >> For the vtable optimization for non-overridden methods I don't see a >> possibility to do this at all. If we compile everything into a library we >> never know when a class of the application overrides something in the JDK, >> etc. >> >> For the selector coloring we could implement a possibility to save the end >> state of the colored graph for the library and use this information when we >> color the selector graph of the application. >> >> But I agree with Sascha: it would be nice to know if this would speed >> things up significantly before we start with something like that. >> >> Markus >> >> 2011/6/10 Sascha Haeberling < <sa...@gm...>sa...@gm...> >> >>> Are we even sure that linking against such a huge library would really be >>> that much faster? From my experience linking huge binaries can take a long >>> time as well. >>> >>> In the end it looks like this: timeSaved = (timePipelineNow - >>> timePipelineThen) - timeLinking >>> >>> If the times saved is small, the question is whether it makes sense >>> investing time into it at all. Therefore it would be good to have some >>> estimates first. >>> >>> // Sascha >>> >>> >>> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >>> >>>> Yes, we've had this discussion before. Basically creating a Unix-like >>>> library (.a) that speeds things compilation time. However, the problem >>>> is that (at least) for now it wouldn't work with the C backend. We do >>>> optimizations that go beyond the simple decision "should this file be >>>> included or not" (what the linker typically does, and which, btw, is >>>> also performed by XMLVM at compile-time). We also optimize things like >>>> vtables, e.g., if a method never gets overridden, it will not end up >>>> in the vtable. This knowledge, however, only exists if we take the >>>> application into account. If the application overrides a library >>>> method, then the cross-compilation of the library also needs to >>>> change. >>>> >>>> I agree that for a regular development cycle it would make sense to >>>> have something like a library and only do full optimizations for the >>>> final release. We would need a way to cross-compile without >>>> optimizations. One big problem I see here are interface methods >>>> (because of the selector coloring scheme that is used), but I better >>>> let Markus respond to this since he implemented those optimizations. >>>> >>>> Arno >>>> >>>> >>>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>>> < <pan...@pa...>pan...@pa...> wrote: >>>> >>>> > >>>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>>> > >>>> >> Panayotis, >>>> >> >>>> >> sorry, I didn't understand what you mean. What do you mean by "the >>>> library"? The C backend does compute all dependencies so that the number of >>>> classes that are compiled and outputted in the end is at a minimum. Only >>>> required classes will be pulled in. Why would you want to replace this and >>>> with what? >>>> >> >>>> >> // Sascha >>>> > >>>> > >>>> > This is from an old discussion. >>>> > >>>> > The idea is to compile the whole JDK and iPhone compatibility library >>>> (and probably other dependencies) into a C library (i.e. in a form of .a >>>> file) and compile the project against this library. >>>> > Thus the Xcode project will compile much much faster and even the >>>> production of the Xcode project will be faster (and smaller) since no files >>>> from the JDK and the compatibility library will be required any more. >>>> > >>>> > This idea was also in the earlier days of ObjC backend, but due to the >>>> way ObjC handled dependencies and late binding, the gain was far less. >>>> > Now with the strong typed C language, things should be much better. >>>> > >>>> > For example, if a function is not used at all, the C compiler >>>> understands that and never embeds it into the output binary - we have >>>> optimizations in the method-level, instead of the class-level. >>>> > >>>> > >>>> > I don't know how virtual functions are handled and probably Arno could >>>> help with this. >>>> > >>>> > >>>> > PS: I remember that ProGuard is also able to perform this type of >>>> optimizations and it is already used for this reason in Android projects. >>>> Probably we should also do something similar? >>>> > >>>> ------------------------------------------------------------------------------ >>>> > EditLive Enterprise is the world's most technically advanced content >>>> > authoring tool. Experience the power of Track Changes, Inline Image >>>> > Editing and ensure content is compliant with Accessibility Checking. >>>> > <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>>> > _______________________________________________ >>>> > Xmlvm-developers mailing list >>>> > <Xml...@li...> >>>> Xml...@li... >>>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>>> >>>> ------------------------------------------------------------------------------ >>>> EditLive Enterprise is the world's most technically advanced content >>>> authoring tool. Experience the power of Track Changes, Inline Image >>>> Editing and ensure content is compliant with Accessibility Checking. >>>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>>> _______________________________________________ >>>> Xmlvm-developers mailing list >>>> <Xml...@li...> >>>> Xml...@li... >>>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>>> >>> >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >> > |
|
From: Markus N. <mar...@gm...> - 2011-06-10 16:01:04
|
We would still need to store the selector graph: A class in the app could implement its own interface as well as implement a JDK interface at the same time. For this case I need to know how the original JDK interface was colored to avoid clashes. For the vtable optimization implementing the "development mode" should be simple. 2011/6/10 Arno Puder <ar...@pu...> > > I'm thinking of a "development mode" where no optimizations are performed. > For vtable this means that all methods go into the vtable by default. That > should be much easier than disabling itable optimizations. > > With regards to time savings: I suspect that a library would be beneficial. > Keep in mind that compile time includes XMLVM compile time as well as Xcode > compile time. XMLVM would only need to compile the application without > computing the transitive closure of all referenced classes (let alone cross > compiling these classes) and Xcode would "only" need to link a huge library > (where I suspect that that will be faster than compiling everything). But of > course I don't have empirical evidence. > > Arno > > > On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> > wrote: > > Hi, > > For the vtable optimization for non-overridden methods I don't see a > possibility to do this at all. If we compile everything into a library we > never know when a class of the application overrides something in the JDK, > etc. > > For the selector coloring we could implement a possibility to save the end > state of the colored graph for the library and use this information when we > color the selector graph of the application. > > But I agree with Sascha: it would be nice to know if this would speed > things up significantly before we start with something like that. > > Markus > > 2011/6/10 Sascha Haeberling < <sa...@gm...>sa...@gm...> > >> Are we even sure that linking against such a huge library would really be >> that much faster? From my experience linking huge binaries can take a long >> time as well. >> >> In the end it looks like this: timeSaved = (timePipelineNow - >> timePipelineThen) - timeLinking >> >> If the times saved is small, the question is whether it makes sense >> investing time into it at all. Therefore it would be good to have some >> estimates first. >> >> // Sascha >> >> >> 2011/6/10 Arno Puder < <ar...@pu...>ar...@pu...> >> >>> Yes, we've had this discussion before. Basically creating a Unix-like >>> library (.a) that speeds things compilation time. However, the problem >>> is that (at least) for now it wouldn't work with the C backend. We do >>> optimizations that go beyond the simple decision "should this file be >>> included or not" (what the linker typically does, and which, btw, is >>> also performed by XMLVM at compile-time). We also optimize things like >>> vtables, e.g., if a method never gets overridden, it will not end up >>> in the vtable. This knowledge, however, only exists if we take the >>> application into account. If the application overrides a library >>> method, then the cross-compilation of the library also needs to >>> change. >>> >>> I agree that for a regular development cycle it would make sense to >>> have something like a library and only do full optimizations for the >>> final release. We would need a way to cross-compile without >>> optimizations. One big problem I see here are interface methods >>> (because of the selector coloring scheme that is used), but I better >>> let Markus respond to this since he implemented those optimizations. >>> >>> Arno >>> >>> >>> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >>> < <pan...@pa...>pan...@pa...> wrote: >>> >>> > >>> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >>> > >>> >> Panayotis, >>> >> >>> >> sorry, I didn't understand what you mean. What do you mean by "the >>> library"? The C backend does compute all dependencies so that the number of >>> classes that are compiled and outputted in the end is at a minimum. Only >>> required classes will be pulled in. Why would you want to replace this and >>> with what? >>> >> >>> >> // Sascha >>> > >>> > >>> > This is from an old discussion. >>> > >>> > The idea is to compile the whole JDK and iPhone compatibility library >>> (and probably other dependencies) into a C library (i.e. in a form of .a >>> file) and compile the project against this library. >>> > Thus the Xcode project will compile much much faster and even the >>> production of the Xcode project will be faster (and smaller) since no files >>> from the JDK and the compatibility library will be required any more. >>> > >>> > This idea was also in the earlier days of ObjC backend, but due to the >>> way ObjC handled dependencies and late binding, the gain was far less. >>> > Now with the strong typed C language, things should be much better. >>> > >>> > For example, if a function is not used at all, the C compiler >>> understands that and never embeds it into the output binary - we have >>> optimizations in the method-level, instead of the class-level. >>> > >>> > >>> > I don't know how virtual functions are handled and probably Arno could >>> help with this. >>> > >>> > >>> > PS: I remember that ProGuard is also able to perform this type of >>> optimizations and it is already used for this reason in Android projects. >>> Probably we should also do something similar? >>> > >>> ------------------------------------------------------------------------------ >>> > EditLive Enterprise is the world's most technically advanced content >>> > authoring tool. Experience the power of Track Changes, Inline Image >>> > Editing and ensure content is compliant with Accessibility Checking. >>> > <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> > _______________________________________________ >>> > Xmlvm-developers mailing list >>> > <Xml...@li...> >>> Xml...@li... >>> > <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >>> >>> ------------------------------------------------------------------------------ >>> EditLive Enterprise is the world's most technically advanced content >>> authoring tool. Experience the power of Track Changes, Inline Image >>> Editing and ensure content is compliant with Accessibility Checking. >>> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >>> _______________________________________________ >>> Xmlvm-developers mailing list >>> <Xml...@li...> >>> Xml...@li... >>> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >>> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >>> >> >> >> >> ------------------------------------------------------------------------------ >> EditLive Enterprise is the world's most technically advanced content >> authoring tool. Experience the power of Track Changes, Inline Image >> Editing and ensure content is compliant with Accessibility Checking. >> <http://p.sf.net/sfu/ephox-dev2dev>http://p.sf.net/sfu/ephox-dev2dev >> _______________________________________________ >> Xmlvm-developers mailing list >> <Xml...@li...> >> Xml...@li... >> <https://lists.sourceforge.net/lists/listinfo/xmlvm-developers> >> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> >> > |
|
From: Arno P. <ar...@pu...> - 2011-06-10 15:58:08
|
I'm thinking of a "development mode" where no optimizations are performed. For vtable this means that all methods go into the vtable by default. That should be much easier than disabling itable optimizations. With regards to time savings: I suspect that a library would be beneficial. Keep in mind that compile time includes XMLVM compile time as well as Xcode compile time. XMLVM would only need to compile the application without computing the transitive closure of all referenced classes (let alone cross compiling these classes) and Xcode would "only" need to link a huge library (where I suspect that that will be faster than compiling everything). But of course I don't have empirical evidence. Arno On Jun 10, 2011, at 9:21 AM, Markus Neubrand <mar...@gm...> wrote: Hi, For the vtable optimization for non-overridden methods I don't see a possibility to do this at all. If we compile everything into a library we never know when a class of the application overrides something in the JDK, etc. For the selector coloring we could implement a possibility to save the end state of the colored graph for the library and use this information when we color the selector graph of the application. But I agree with Sascha: it would be nice to know if this would speed things up significantly before we start with something like that. Markus 2011/6/10 Sascha Haeberling <sa...@gm...> > Are we even sure that linking against such a huge library would really be > that much faster? From my experience linking huge binaries can take a long > time as well. > > In the end it looks like this: timeSaved = (timePipelineNow - > timePipelineThen) - timeLinking > > If the times saved is small, the question is whether it makes sense > investing time into it at all. Therefore it would be good to have some > estimates first. > > // Sascha > > > 2011/6/10 Arno Puder <ar...@pu...> > >> Yes, we've had this discussion before. Basically creating a Unix-like >> library (.a) that speeds things compilation time. However, the problem >> is that (at least) for now it wouldn't work with the C backend. We do >> optimizations that go beyond the simple decision "should this file be >> included or not" (what the linker typically does, and which, btw, is >> also performed by XMLVM at compile-time). We also optimize things like >> vtables, e.g., if a method never gets overridden, it will not end up >> in the vtable. This knowledge, however, only exists if we take the >> application into account. If the application overrides a library >> method, then the cross-compilation of the library also needs to >> change. >> >> I agree that for a regular development cycle it would make sense to >> have something like a library and only do full optimizations for the >> final release. We would need a way to cross-compile without >> optimizations. One big problem I see here are interface methods >> (because of the selector coloring scheme that is used), but I better >> let Markus respond to this since he implemented those optimizations. >> >> Arno >> >> >> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >> <pan...@pa...> wrote: >> >> > >> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >> > >> >> Panayotis, >> >> >> >> sorry, I didn't understand what you mean. What do you mean by "the >> library"? The C backend does compute all dependencies so that the number of >> classes that are compiled and outputted in the end is at a minimum. Only >> required classes will be pulled in. Why would you want to replace this and >> with what? >> >> >> >> // Sascha >> > >> > >> > This is from an old discussion. >> > >> > The idea is to compile the whole JDK and iPhone compatibility library >> (and probably other dependencies) into a C library (i.e. in a form of .a >> file) and compile the project against this library. >> > Thus the Xcode project will compile much much faster and even the >> production of the Xcode project will be faster (and smaller) since no files >> from the JDK and the compatibility library will be required any more. >> > >> > This idea was also in the earlier days of ObjC backend, but due to the >> way ObjC handled dependencies and late binding, the gain was far less. >> > Now with the strong typed C language, things should be much better. >> > >> > For example, if a function is not used at all, the C compiler >> understands that and never embeds it into the output binary - we have >> optimizations in the method-level, instead of the class-level. >> > >> > >> > I don't know how virtual functions are handled and probably Arno could >> help with this. >> > >> > >> > PS: I remember that ProGuard is also able to perform this type of >> optimizations and it is already used for this reason in Android projects. >> Probably we should also do something similar? >> > >> ------------------------------------------------------------------------------ >> > EditLive Enterprise is the world's most technically advanced content >> > authoring tool. Experience the power of Track Changes, Inline Image >> > Editing and ensure content is compliant with Accessibility Checking. >> > http://p.sf.net/sfu/ephox-dev2dev >> > _______________________________________________ >> > Xmlvm-developers mailing list >> > Xml...@li... >> > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> >> >> ------------------------------------------------------------------------------ >> EditLive Enterprise is the world's most technically advanced content >> authoring tool. Experience the power of Track Changes, Inline Image >> Editing and ensure content is compliant with Accessibility Checking. >> http://p.sf.net/sfu/ephox-dev2dev >> _______________________________________________ >> Xmlvm-developers mailing list >> Xml...@li... >> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> > > > > ------------------------------------------------------------------------------ > EditLive Enterprise is the world's most technically advanced content > authoring tool. Experience the power of Track Changes, Inline Image > Editing and ensure content is compliant with Accessibility Checking. > http://p.sf.net/sfu/ephox-dev2dev > _______________________________________________ > Xmlvm-developers mailing list > Xml...@li... > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers > > |
|
From: Markus N. <mar...@gm...> - 2011-06-10 15:21:12
|
Hi, For the vtable optimization for non-overridden methods I don't see a possibility to do this at all. If we compile everything into a library we never know when a class of the application overrides something in the JDK, etc. For the selector coloring we could implement a possibility to save the end state of the colored graph for the library and use this information when we color the selector graph of the application. But I agree with Sascha: it would be nice to know if this would speed things up significantly before we start with something like that. Markus 2011/6/10 Sascha Haeberling <sa...@gm...> > Are we even sure that linking against such a huge library would really be > that much faster? From my experience linking huge binaries can take a long > time as well. > > In the end it looks like this: timeSaved = (timePipelineNow - > timePipelineThen) - timeLinking > > If the times saved is small, the question is whether it makes sense > investing time into it at all. Therefore it would be good to have some > estimates first. > > // Sascha > > > 2011/6/10 Arno Puder <ar...@pu...> > >> Yes, we've had this discussion before. Basically creating a Unix-like >> library (.a) that speeds things compilation time. However, the problem >> is that (at least) for now it wouldn't work with the C backend. We do >> optimizations that go beyond the simple decision "should this file be >> included or not" (what the linker typically does, and which, btw, is >> also performed by XMLVM at compile-time). We also optimize things like >> vtables, e.g., if a method never gets overridden, it will not end up >> in the vtable. This knowledge, however, only exists if we take the >> application into account. If the application overrides a library >> method, then the cross-compilation of the library also needs to >> change. >> >> I agree that for a regular development cycle it would make sense to >> have something like a library and only do full optimizations for the >> final release. We would need a way to cross-compile without >> optimizations. One big problem I see here are interface methods >> (because of the selector coloring scheme that is used), but I better >> let Markus respond to this since he implemented those optimizations. >> >> Arno >> >> >> On Jun 10, 2011, at 6:26 AM, Panayotis Katsaloulis >> <pan...@pa...> wrote: >> >> > >> > On 10 Ιουν 2011, at 3:11 μ.μ., Sascha Haeberling wrote: >> > >> >> Panayotis, >> >> >> >> sorry, I didn't understand what you mean. What do you mean by "the >> library"? The C backend does compute all dependencies so that the number of >> classes that are compiled and outputted in the end is at a minimum. Only >> required classes will be pulled in. Why would you want to replace this and >> with what? >> >> >> >> // Sascha >> > >> > >> > This is from an old discussion. >> > >> > The idea is to compile the whole JDK and iPhone compatibility library >> (and probably other dependencies) into a C library (i.e. in a form of .a >> file) and compile the project against this library. >> > Thus the Xcode project will compile much much faster and even the >> production of the Xcode project will be faster (and smaller) since no files >> from the JDK and the compatibility library will be required any more. >> > >> > This idea was also in the earlier days of ObjC backend, but due to the >> way ObjC handled dependencies and late binding, the gain was far less. >> > Now with the strong typed C language, things should be much better. >> > >> > For example, if a function is not used at all, the C compiler >> understands that and never embeds it into the output binary - we have >> optimizations in the method-level, instead of the class-level. >> > >> > >> > I don't know how virtual functions are handled and probably Arno could >> help with this. >> > >> > >> > PS: I remember that ProGuard is also able to perform this type of >> optimizations and it is already used for this reason in Android projects. >> Probably we should also do something similar? >> > >> ------------------------------------------------------------------------------ >> > EditLive Enterprise is the world's most technically advanced content >> > authoring tool. Experience the power of Track Changes, Inline Image >> > Editing and ensure content is compliant with Accessibility Checking. >> > http://p.sf.net/sfu/ephox-dev2dev >> > _______________________________________________ >> > Xmlvm-developers mailing list >> > Xml...@li... >> > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> >> >> ------------------------------------------------------------------------------ >> EditLive Enterprise is the world's most technically advanced content >> authoring tool. Experience the power of Track Changes, Inline Image >> Editing and ensure content is compliant with Accessibility Checking. >> http://p.sf.net/sfu/ephox-dev2dev >> _______________________________________________ >> Xmlvm-developers mailing list >> Xml...@li... >> https://lists.sourceforge.net/lists/listinfo/xmlvm-developers >> > > > > ------------------------------------------------------------------------------ > EditLive Enterprise is the world's most technically advanced content > authoring tool. Experience the power of Track Changes, Inline Image > Editing and ensure content is compliant with Accessibility Checking. > http://p.sf.net/sfu/ephox-dev2dev > _______________________________________________ > Xmlvm-developers mailing list > Xml...@li... > https://lists.sourceforge.net/lists/listinfo/xmlvm-developers > > |
