From: Sean P. <sea...@ma...> - 2012-11-06 23:15:50
|
I've started what is planned to be a long migration of ASL to github. The ASL perforce database is now locked. You can find the current ASL mainline sources at: github/stlab/legacy The reason why they are in the legacy repo is that I plan to move the libraries to a structure conforming with the boost modular libraries (github/boost-lib). I will be starting with some of the libraries I've been using in a product I've been working on at Adobe (currently these are forked from ASL, so I'm doing this as part of upstreaming that work). Beside moving to github and module based libraries, I'm also working on: Updating ASL to C++11, removing unnecessary dependencies from boost (where C++11 libraries suffice) and removing overlap with newer boost libraries. Simplifying ASL by removing the ABI stable support for those libraries that use it, and taking a more pragmatic approach to the libraries (many were done as experiments where the complexity is unwarranted given the results). As modules are split out and cleaned up I plan to propose them for inclusion in boost as appropriate. I'd also like to consolidate as much of the remaining ASL infrastructure (we currently have mailing lists and issue tracking on source forge, website and wiki hosted by Adobe, and now source code moving to github). Suggestions and recommendations welcome. I'm hoping that moving to github will make it easier to collaborate with others on ASL. If you would like to help, let me know. If you have suggestions, let me know. Thanks, Sean |
From: Sean P. <sea...@ma...> - 2012-11-07 00:37:09
|
I setup a section on our wiki to document the transition to github. That can be found here: http://stlab.adobe.com/wiki/index.php/Github_migration Not a lot there at the moment - but should be enough to get people started. Sean On Tue, Nov 6, 2012 at 3:15 PM, Sean Parent <sea...@ma...> wrote: > I've started what is planned to be a long migration of ASL to github. The > ASL perforce database is now locked. > > You can find the current ASL mainline sources at: > > github/stlab/legacy > > The reason why they are in the legacy repo is that I plan to move the > libraries to a structure conforming with the boost modular libraries > (github/boost-lib). > > I will be starting with some of the libraries I've been using in a product > I've been working on at Adobe (currently these are forked from ASL, so I'm > doing this as part of upstreaming that work). > > Beside moving to github and module based libraries, I'm also working on: > > Updating ASL to C++11, removing unnecessary dependencies from boost (where > C++11 libraries suffice) and removing overlap with newer boost libraries. > > Simplifying ASL by removing the ABI stable support for those libraries > that use it, and taking a more pragmatic approach to the libraries (many > were done as experiments where the complexity is unwarranted given the > results). > > As modules are split out and cleaned up I plan to propose them for > inclusion in boost as appropriate. > > I'd also like to consolidate as much of the remaining > ASL infrastructure (we currently have mailing lists and issue tracking on > source forge, website and wiki hosted by Adobe, and now source code moving > to github). Suggestions and recommendations welcome. > > I'm hoping that moving to github will make it easier to collaborate with > others on ASL. If you would like to help, let me know. If you have > suggestions, let me know. > > Thanks, > Sean > |
From: Ivan Le L. <iva...@fr...> - 2012-11-07 22:02:02
|
----- Mail original ----- > De: "Sean Parent" <sea...@ma...> > À: ado...@li..., "Mat Marcus" > <mm...@em...>, "Foster Brereton" <fbr...@ad...> > Envoyé: Mercredi 7 Novembre 2012 01:37:03 > Objet: Re: [Adobe-source-devel] Moving ASL to github > I setup a section on our wiki to document the transition to github. > That can be found here: > http://stlab.adobe.com/wiki/index.php/Github_migration > Not a lot there at the moment - but should be enough to get people > started. Thank you for doing this. I will definitely base my (utterly slow) work on ASL on this repository. As I do not understand anything to Perforce, merging with .43 release has been somewhat painful. > > As modules are split out and cleaned up I plan to propose them for > > inclusion in boost as appropriate. > Does that mean that forking is not recommended at the moment and until modules are split ? Regards, Ivan |
From: Sean P. <sea...@ma...> - 2012-11-07 23:02:48
|
Feel free to fork - splitting into modules is going to take some time. ----- First up is any_regular.hpp Issues with the current implementation to be addressed: 1) reliance on promote<> - promote<> seemed like a good idea but it has bit me a number of times, currently longs promote to double, which is just wrong. It also means that the interface needs to be value based in many locations where it could have been reference based. Will add a build flag to tag places where promote would have kicked in to help with migration. 2) needs to be updated to C++11 move, and remove dependency on move library. 3) assignment - currently requires assignment, this is an issue because I'd like to make function objects and lambda's first class (or as close as possible) but equality is very useful. I have been playing with ways to write a has_equal_to<> that seems to work good enough. 4) implementation and interface complicated by ABI stability. ABI stability to be removed (but I will be moving to using inlined namespaces). 5) implementation complicated by use of poly<> - plan to remove that. 6) rename to adobe::any 7) make constructor implicit 8) rename explicit_cast and runtime_cast to static_cast_ and dynamic_cast_ to make them easier to remember. Open issues: 1) Match boost::any api including broken any_cast interface? (differs from boost::any by support small object optimization, optional equal_to, and unsafe static cast). boost::any has been proposed to the standard - and the broken any_cast has already been discussed to be fixed. 2) Add support for other operators - such as index by std::size_t or const char* (useful for dictionaries and arrays), arithmetic operators for numeric types? Suggestions? Comments? Sean On Wed, Nov 7, 2012 at 2:01 PM, Ivan Le Lann <iva...@fr...> wrote: > > > ------------------------------ > > *De: *"Sean Parent" <sea...@ma...> > *À: *ado...@li..., "Mat Marcus" < > mm...@em...>, "Foster Brereton" <fbr...@ad...> > *Envoyé: *Mercredi 7 Novembre 2012 01:37:03 > *Objet: *Re: [Adobe-source-devel] Moving ASL to github > > > I setup a section on our wiki to document the transition to github. That > can be found here: > > http://stlab.adobe.com/wiki/index.php/Github_migration > > Not a lot there at the moment - but should be enough to get people started. > > > Thank you for doing this. > I will definitely base my (utterly slow) work on ASL on this repository. > As I do not understand anything to Perforce, merging with .43 release has > been somewhat painful. > > > > >> As modules are split out and cleaned up I plan to propose them for >> inclusion in boost as appropriate. >> >> > > Does that mean that forking is not recommended at the moment and until > modules are split ? > > Regards, > Ivan > > > |
From: Sean P. <sea...@ma...> - 2012-11-09 04:51:35
|
I've move the migration page on the wiki < http://stlab.adobe.com/wiki/index.php/GitHub_Migration_and_Status> and updated the status. Assistance from anyone with experience in bjam and/or cmake would be greately appreciated at this point (see the status). Sean On Wed, Nov 7, 2012 at 3:02 PM, Sean Parent <sea...@ma...> wrote: > Feel free to fork - splitting into modules is going to take some time. > > ----- > > First up is any_regular.hpp > Issues with the current implementation to be addressed: > 1) reliance on promote<> - promote<> seemed like a good idea but it > has bit me a number of times, currently longs promote to double, which is > just wrong. It also means that the interface needs to be value based in > many locations where it could have been reference based. Will add a build > flag to tag places where promote would have kicked in to help with > migration. > 2) needs to be updated to C++11 move, and remove dependency on move > library. > 3) assignment - currently requires assignment, this is an issue > because I'd like to make function objects and lambda's first class (or as > close as possible) but equality is very useful. I have been playing with > ways to write a has_equal_to<> that seems to work good enough. > 4) implementation and interface complicated by ABI stability. ABI > stability to be removed (but I will be moving to using inlined namespaces). > 5) implementation complicated by use of poly<> - plan to remove > that. > 6) rename to adobe::any > 7) make constructor implicit > 8) rename explicit_cast and runtime_cast to static_cast_ and > dynamic_cast_ to make them easier to remember. > > Open issues: > 1) Match boost::any api including broken any_cast interface? > (differs from boost::any by support small object optimization, optional > equal_to, and unsafe static cast). boost::any has been proposed to the > standard - and the broken any_cast has already been discussed to be fixed. > 2) Add support for other operators - such as index by std::size_t > or const char* (useful for dictionaries and arrays), arithmetic operators > for numeric types? > > Suggestions? Comments? > Sean > > > On Wed, Nov 7, 2012 at 2:01 PM, Ivan Le Lann <iva...@fr...> wrote: > >> >> >> ------------------------------ >> >> *De: *"Sean Parent" <sea...@ma...> >> *À: *ado...@li..., "Mat Marcus" < >> mm...@em...>, "Foster Brereton" <fbr...@ad...> >> *Envoyé: *Mercredi 7 Novembre 2012 01:37:03 >> *Objet: *Re: [Adobe-source-devel] Moving ASL to github >> >> >> I setup a section on our wiki to document the transition to github. That >> can be found here: >> >> http://stlab.adobe.com/wiki/index.php/Github_migration >> >> Not a lot there at the moment - but should be enough to get people >> started. >> >> >> Thank you for doing this. >> I will definitely base my (utterly slow) work on ASL on this repository. >> As I do not understand anything to Perforce, merging with .43 release has >> been somewhat painful. >> >> >> >> >>> As modules are split out and cleaned up I plan to propose them for >>> inclusion in boost as appropriate. >>> >>> >> >> Does that mean that forking is not recommended at the moment and until >> modules are split ? >> >> Regards, >> Ivan >> >> >> > |
From: Foster B. <fbr...@ad...> - 2012-11-15 18:52:07
|
I have taken a stab at this and made notes in the legacy/development branch accordingly. The BOOST_NO_CXX11_NUMERIC_LIMITS macro not being defined seems to be a Clang support issue within boost/config. It's being defined appropriately when the toolset is darwin (gcc-4.2.1) Note: Using Apple's dev tools using clang is now a requirement given the use of c++11 features that only exist in gcc versions 4.4 or later. Xcode 4.5.2 uses a flavor of gcc 4.2.1. I'm sure Sean and Mat are aware of this, but I wanted to call it out explicitly. Of course if you're building with a version of gcc that supports the c++11 features used, you're in the clear. With bjam I haven't been able to figure out a way to force the entire code base of Begin to build with the x86 architecture. The command line I've been using is: bjam -a release -j8 address-model=32 Note: -a is a force-rebuild of all sources and is optional. -j8 specifies 8 concurrent compilation processes for my 8-core machine; YMMV. Blessings, Foster -- Foster T. Brereton ΙΧΘΥΣ Ro.3:21-26 Senior Computer Scientist --- Photoshop Engineering --- Adobe On 11/8/12 8:51p, "Sean Parent" <sea...@ma...> wrote: >I've move the migration page on the wiki ><http://stlab.adobe.com/wiki/index.php/GitHub_Migration_and_Status> and >updated the status. > >Assistance from anyone with experience in bjam and/or cmake would be >greately appreciated at this point (see the status). > >Sean > > >On Wed, Nov 7, 2012 at 3:02 PM, Sean Parent <sea...@ma...> wrote: > >Feel free to fork - splitting into modules is going to take some time. >----- > >First up is any_regular.hpp > Issues with the current implementation to be addressed: > 1) reliance on promote<> - promote<> seemed like a good idea but >it has bit me a number of times, currently longs promote to double, which >is just wrong. It also means that the interface needs to be value based >in many locations where it could have been reference based. Will add a >build flag to tag places where promote would have kicked in to help with >migration. > 2) needs to be updated to C++11 move, and remove dependency on >move library. > 3) assignment - currently requires assignment, this is an issue >because I'd like to make function objects and lambda's first class (or as >close as possible) but equality is very useful. I have been playing with >ways to write a has_equal_to<> that seems to work good enough. > 4) implementation and interface complicated by ABI stability. ABI >stability to be removed (but I will be moving to using inlined >namespaces). > 5) implementation complicated by use of poly<> - plan to remove >that. > 6) rename to adobe::any > 7) make constructor implicit > 8) rename explicit_cast and runtime_cast to static_cast_ and >dynamic_cast_ to make them easier to remember. > > >Open issues: > 1) Match boost::any api including broken any_cast interface? >(differs from boost::any by support small object optimization, optional >equal_to, and unsafe static cast). boost::any has been proposed to the >standard - and the broken any_cast has already been discussed to be fixed. > 2) Add support for other operators - such as index by std::size_t >or const char* (useful for dictionaries and arrays), arithmetic operators >for numeric types? > >Suggestions? Comments? >Sean > > > >On Wed, Nov 7, 2012 at 2:01 PM, Ivan Le Lann <iva...@fr...> wrote: > > > >________________________________________ > >De: "Sean Parent" <sea...@ma...> >À: ado...@li..., "Mat Marcus" ><mm...@em...>, "Foster Brereton" <fbr...@ad...> >Envoyé: Mercredi 7 Novembre 2012 01:37:03 >Objet: Re: [Adobe-source-devel] Moving ASL to github > >I setup a section on our wiki to document the transition to github. That >can be found here: > >http://stlab.adobe.com/wiki/index.php/Github_migration > > >Not a lot there at the moment - but should be enough to get people >started. > > > > >Thank you for doing this. >I will definitely base my (utterly slow) work on ASL on this repository. > >As I do not understand anything to Perforce, merging with .43 release has >been somewhat painful. > > > > > >As modules are split out and cleaned up I plan to propose them for >inclusion in boost as appropriate. > > > > > > > > > > >Does that mean that forking is not recommended at the moment and until >modules are split ? > >Regards, >Ivan > > > > > > > > > > > > > > > > |