From: Mattia B. <mat...@li...> - 2007-02-17 15:16:58
|
On Tue, 13 Feb 2007 14:00:35 -0800 "Donovan Allen" <al...@Zo...> wrote: > ________________________________ > > From: Mark Dootson [mailto:mar...@zn...] > Sent: Mon 2/12/2007 6:45 PM > To: Donovan Allen > Cc: wxp...@li... > Subject: Re: [wxperl-users] wxPerl Community project... > > > I guess it goes back to Huub's original mail in that a common repository > > for all things wxPerl would be quite nice. I have no real opinion about > > what the back end to this should be nor where it might be hosted other > > than that as it all ultimately depends on Mattias' work, you'd need a > > good reason not to host it at wxperl.sourceforge.net. I'm not saying > > there aren't good reasons - its just a straightforward non-loaded statement. I would be happy to give access to SF web site to anybody who is willing to add/improve/maintain it. > I don't have a stronge opion here. Not sure what I said that was or wasn't in support of this. However, whatever the hosting solution is, it must be maintable by more than one person. I think that was one aspect of Huub's original concept and also why there are other wxPerl wiki's out there...the need for a community to be in control of it's own destiny. Regards Mattia |
From: Donovan A. <al...@Zo...> - 2007-02-13 22:10:28
|
>> I guess it goes back to Huub's original mail in that a common = repository >> for all things wxPerl would be quite nice. I have no real opinion = about >> what the back end to this should be nor where it might be hosted = other >> than that as it all ultimately depends on Mattias' work, you'd need a >> good reason not to host it at wxperl.sourceforge.net. I'm not saying >> there aren't good reasons - its just a straightforward non-loaded = statement. > I don't have a stronge opion here. Not sure what I said that was or = wasn't in support of this. However, whatever the hosting solution is, = it must be maintable by =20 > more than one person. I think that was one aspect of Huub's original = concept and also why there are other wxPerl wiki's out there...the need = for a community to be=20 > in control of it's own destiny. I forgot to add, that without software admin rights, I also think that = random wiki's on other people's servers does not in fact satisfy the = larger need of community maintainability. =20 ...and egad...forgive the spelling and typos... =20 |
From: Mike S. <Mik...@Do...> - 2007-02-13 01:01:49
|
Eric Wilhelm wrote: > BTW, the TAPx project would love to have a wxPerl gui test harness > example. There's something that could be used by many perl > programmers, even programmers of other languages, and has a small, but > non-trivial issue of IPC with running children in a fork. IMO, it > would also be a great place to apply a Wx::TreeListCtrl widget. > Oh -- we've had another contractor (Brad Embree) working the past few months on a test harness for wxPerl. At this point it assumes you will be using the Wx::Data controls (see other email) which are "test-aware". You can "record" a test, which gets saved (currently using YAML I think??). You can then "replay" a test and the test harness will compare actual vs expected. I'll see if Brad can write up something more meaningful about this. Mike. |
From: Eric W. <scr...@gm...> - 2007-02-13 05:07:21
|
# from Bradley K. Embree # on Monday 12 February 2007 08:37 pm: >Mike Schroeder wrote: >> Eric Wilhelm wrote: >>> BTW, the TAPx project would love to have a wxPerl gui test harness >>> example. There's something that could be used by many perl >>> programmers, even programmers of other languages, and has a small, >>> but non-trivial issue of IPC with running children in a fork. IMO, >>> it would also be a great place to apply a Wx::TreeListCtrl widget. >> >> Oh -- we've had another contractor (Brad Embree) working the past >> few months on a test harness for wxPerl. At this point it assumes >I am actually in the process of finishing the test harness up, and >integrating it back into our main trunk for some final testing. That sounds really neat, but we need a jargon clarification: TAP Harness - thing that runs tests and aggregates results TAP producer - Testing framework (e.g. Test::More) So, if you're code prints "ok", "not ok" and involves .t files run under "make test", then it is a testing framework (TAP producer.) If it is actually a harness, that makes it difficult to run within a harness. A harness (e.g. `prove` or `runtests`) simply takes a list of files to test and executes each one, reporting on what passed/failed and giving a summary at the end. What the TAPx project would like is a wxPerl test harness (i.e. wxprove) which runs your tests (be they gui tests or not) and shows which ones passed or failed. Yes, we want and need a wxPerl test framework too. :-D However, the widespread and cross-language usefulness of which I wrote only applies to a harness. --Eric -- A counterintuitive sansevieria trifasciata was once literalized guiltily. --Product of Artificial Intelligence --------------------------------------------------- http://scratchcomputing.com --------------------------------------------------- |
From: Bradley K. E. <Bra...@Do...> - 2007-02-13 06:09:28
|
Eric Wilhelm wrote: > # from Bradley K. Embree > # on Monday 12 February 2007 08:37 pm: > >> Mike Schroeder wrote: >>> Eric Wilhelm wrote: >>>> BTW, the TAPx project would love to have a wxPerl gui test harness >>>> example. There's something that could be used by many perl >>>> programmers, even programmers of other languages, and has a small, >>>> but non-trivial issue of IPC with running children in a fork. IMO, >>>> it would also be a great place to apply a Wx::TreeListCtrl widget. >>> Oh -- we've had another contractor (Brad Embree) working the past >>> few months on a test harness for wxPerl. At this point it assumes >> I am actually in the process of finishing the test harness up, and >> integrating it back into our main trunk for some final testing. > > That sounds really neat, but we need a jargon clarification: > > TAP Harness - thing that runs tests and aggregates results > > TAP producer - Testing framework (e.g. Test::More) > > So, if you're code prints "ok", "not ok" and involves .t files run under > "make test", then it is a testing framework (TAP producer.) If it is > actually a harness, that makes it difficult to run within a harness. > > A harness (e.g. `prove` or `runtests`) simply takes a list of files to > test and executes each one, reporting on what passed/failed and giving > a summary at the end. > > What the TAPx project would like is a wxPerl test harness (i.e. wxprove) > which runs your tests (be they gui tests or not) and shows which ones > passed or failed. > > Yes, we want and need a wxPerl test framework too. :-D However, the > widespread and cross-language usefulness of which I wrote only applies > to a harness. > > --Eric Agreed, and I think the "playback" of a recorded test session could be made to emit TAP as it plays back the events. I imagine with a little more work we could have playback fire up the app, and run through the events emitting TAP as it goes. I've looked at TAPx a little bit, but the last time I played with it I was bitten by some problems it had on windows. I believe those problems have since been ironed out, and I have been meaning to take another look. Brad |
From: Foo JH <jhf...@ex...> - 2007-02-13 02:49:43
|
Firstly, I'd like to applaud Peters for his proactive stance towards the problem statement. I think there's a (relatively) small but helpful community in this mailing list, who have once faced the pain of learning the wxPerl framework, but have since learnt enough on their own to be less bothered by them. I however am one who sort of got stuck with the pain, and have since resorted to C# to get around the issues. It's a sad departure, and one I'd like to return to again. Contrary to popular belief, programming stuff in Perl is REALLY fast and efficient (but the memory footprint can be improved...). Like any other community, wxPerl needs to be friendly to 'new blood'. And for these people, we need something simple for them to get started on. What we have now are: 1. A couple of aged tutorials from perl.com and sourceforge.net 2. A few sets of integrated demos showing how the GUI components work, with source codes 3. An aged documentation on the classes IMHO, these stuff are good, but need some form of updating to the more recent library. Looking at how 'others have done it', a constant (if even a trickle) stream of tips and tricks will be very encouraging, especially for people who need to know if wxPerl will be worth their time and effort. Let's face it. Perl is really starting to look like Cobol in the sense of mindshare of the developer community. Slashdot ppl have been questioning the relavance of Perl in today's OS environments. And it's not because Perl sucks, but because they don't see anything new coming from this community. It's a human mentality, and it is a self-fulfilling prophecy. To sum it up, if we can get a few good men to put up a fresh new look for wxPerl, I'm all for it. Just let me know where I can butt in. Huub Peters wrote: > Hello list, > > I know many others have proposed it in the past and (maybe not so) many > initiatives have been actually executed but we really need a central > community kind of place. > > The german wiki (http://www.wxperl.de) is a mess (spammed all the time > and even a defacing at this moment). > The wiki at http://wxperl.pvoice.org/kwiki/ is a nice effort but only > thanks to some users (or even only one? jralls?) it is staying free of > spam but there are no new articles posted at all. Why? Don't know. > It's probably not known and personaly I just don't like the Kwiki-software > but that's a matter of taste off course. > (On a side note: Jouke (whose place this wiki is hosted on) has anounced > on his weblog on 24-09-2006 that he will be offline for an undetermined > ammount of time due to personal issues. Otherwise I would have taken this > up with him directly to begin with.) > > There are other spots with excelent stuff (like gigi.co.uk) but I only heard > about that on this list and it is decentralized this way. > > I can go on with being negative and summing up what I think is "wrong" > and compare to "the great wxPython community" but that won't solve a thing > so here's what I came up with: > > * I took the liberty of registering wxperl.info. > A good domain name improves "findability" > * I'm in the lucky position of having hosting facilities and a lot of > fast bandwidth to go with it that I can use. (the company where I work > has no problems with me using it) > * I started writing this mail :-) > > So what's next...? > Next I need ideas and help. > > Here are some of my ideas: > * We need a user friendly, nice looking wiki. > Something like "Mediawiki" (the heart of Wikipedia) but offcourse I > can't live with a perl community on a php-driven site so after some > searching I found Twiki: http://www.twiki.org which looks robust enough > to me. > * We need tutorials > * We need Howto's. > * We need tips and tricks. > * Maybe write up the problems and solutions from the mailing list. > * Maybe a forum for some nice discussions (alltough it might clash with > the mailing list) > * Show of your work. > > Bottom line: we need a central community where beginners and experts can > teach and learn about wxPerl. > I'm trying to offer the place to start this and hope it will grow to be as > big as > wxPerl deserves it to be. > > E-mail me on or off list with your thoughts/ideas/offers etcetera and > maybe we'll have a nice community in no time. > > Huub Peters > > > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job easier. > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > wxperl-users mailing list > wxp...@li... > https://lists.sourceforge.net/lists/listinfo/wxperl-users > |
From: Octavian R. <ora...@gm...> - 2007-02-13 07:12:31
|
Hi, I am less than a beginner in WXPerl, and maybe it will help you if I tell you what stopped me learning WXPerl. I have tried first learning WXPerl, (before using Win32::GUI) and I didn't like its syntax. The Win32::GUI syntax is much clear. Well I think the syntax cannot be changed, but maybe it could be better explained (like for those who don't know C or other languages than perl). Then I have tried some WXPerl demos and here I found many issues. There are many widgets that are not accessible for everyone (I am blind and I am using a screen reader). For example, the HTML widget is not accessible. Maybe some widgets depend on WX widgets and cannot be made accessible in WXPerl, but I think all the demos should be as much accessible as possible only using a keyboard (with no mouse). Regarding the accessibility, Win32::GUI is much better accessible. I became interested again in WXPerl, because Win32::GUI doesn't support UTF-8 and I need this, but I still find it same inaccessible and hard to learn. I have tried reading the WXPerl manual (that in .chm format), but I have seen that for very many widgets and methods "this is not implemented in WXPerl yet" and this is not very encouraging. And I have also seen that WXPython is better promoted. (Too bad, because I don't know Python). Octavian ----- Original Message ----- From: "Foo JH" <jhf...@ex...> To: "Huub Peters" <hu...@tr...> Cc: <wxp...@li...> Sent: Tuesday, February 13, 2007 4:49 AM Subject: Re: [wxperl-users] wxPerl Community project... > Firstly, I'd like to applaud Peters for his proactive stance towards the > problem statement. > > I think there's a (relatively) small but helpful community in this > mailing list, who have once faced the pain of learning the wxPerl > framework, but have since learnt enough on their own to be less bothered > by them. > > I however am one who sort of got stuck with the pain, and have since > resorted to C# to get around the issues. It's a sad departure, and one > I'd like to return to again. Contrary to popular belief, programming > stuff in Perl is REALLY fast and efficient (but the memory footprint can > be improved...). > > Like any other community, wxPerl needs to be friendly to 'new blood'. > And for these people, we need something simple for them to get started > on. What we have now are: > 1. A couple of aged tutorials from perl.com and sourceforge.net > 2. A few sets of integrated demos showing how the GUI components work, > with source codes > 3. An aged documentation on the classes > > IMHO, these stuff are good, but need some form of updating to the more > recent library. Looking at how 'others have done it', a constant (if > even a trickle) stream of tips and tricks will be very encouraging, > especially for people who need to know if wxPerl will be worth their > time and effort. > > Let's face it. Perl is really starting to look like Cobol in the sense > of mindshare of the developer community. Slashdot ppl have been > questioning the relavance of Perl in today's OS environments. And it's > not because Perl sucks, but because they don't see anything new coming > from this community. It's a human mentality, and it is a self-fulfilling > prophecy. > > To sum it up, if we can get a few good men to put up a fresh new look > for wxPerl, I'm all for it. Just let me know where I can butt in. > > > > > Huub Peters wrote: >> Hello list, >> >> I know many others have proposed it in the past and (maybe not so) many >> initiatives have been actually executed but we really need a central >> community kind of place. >> >> The german wiki (http://www.wxperl.de) is a mess (spammed all the time >> and even a defacing at this moment). >> The wiki at http://wxperl.pvoice.org/kwiki/ is a nice effort but only >> thanks to some users (or even only one? jralls?) it is staying free of >> spam but there are no new articles posted at all. Why? Don't know. >> It's probably not known and personaly I just don't like the >> Kwiki-software >> but that's a matter of taste off course. >> (On a side note: Jouke (whose place this wiki is hosted on) has anounced >> on his weblog on 24-09-2006 that he will be offline for an undetermined >> ammount of time due to personal issues. Otherwise I would have taken this >> up with him directly to begin with.) >> >> There are other spots with excelent stuff (like gigi.co.uk) but I only >> heard >> about that on this list and it is decentralized this way. >> >> I can go on with being negative and summing up what I think is "wrong" >> and compare to "the great wxPython community" but that won't solve a >> thing >> so here's what I came up with: >> >> * I took the liberty of registering wxperl.info. >> A good domain name improves "findability" >> * I'm in the lucky position of having hosting facilities and a lot of >> fast bandwidth to go with it that I can use. (the company where I work >> has no problems with me using it) >> * I started writing this mail :-) >> >> So what's next...? >> Next I need ideas and help. >> >> Here are some of my ideas: >> * We need a user friendly, nice looking wiki. >> Something like "Mediawiki" (the heart of Wikipedia) but offcourse I >> can't live with a perl community on a php-driven site so after some >> searching I found Twiki: http://www.twiki.org which looks robust >> enough >> to me. >> * We need tutorials >> * We need Howto's. >> * We need tips and tricks. >> * Maybe write up the problems and solutions from the mailing list. >> * Maybe a forum for some nice discussions (alltough it might clash with >> the mailing list) >> * Show of your work. >> >> Bottom line: we need a central community where beginners and experts can >> teach and learn about wxPerl. >> I'm trying to offer the place to start this and hope it will grow to be >> as >> big as >> wxPerl deserves it to be. >> >> E-mail me on or off list with your thoughts/ideas/offers etcetera and >> maybe we'll have a nice community in no time. >> >> Huub Peters >> >> >> >> >> ------------------------------------------------------------------------- >> Using Tomcat but need to do more? Need to support web services, security? >> Get stuff done quickly with pre-integrated technology to make your job >> easier. >> Download IBM WebSphere Application Server v.1.0.1 based on Apache >> Geronimo >> http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 >> _______________________________________________ >> wxperl-users mailing list >> wxp...@li... >> https://lists.sourceforge.net/lists/listinfo/wxperl-users >> > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job > easier. > Download IBM WebSphere Application Server v.1.0.1 based on Apache Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > wxperl-users mailing list > wxp...@li... > https://lists.sourceforge.net/lists/listinfo/wxperl-users |
From: Foo JH <jhf...@ex...> - 2007-02-13 07:24:34
|
This is what I mean: an aging population. If we can't make it easy for newbies to accept the framework, then it becomes a dying art. Nevermind whether it is a better alternative. Octavian Rasnita wrote: > Hi, > > I am less than a beginner in WXPerl, and maybe it will help you if I tell > you what stopped me learning WXPerl. |
From: Bradley K. E. <Bra...@Do...> - 2007-02-13 04:36:59
|
Mike Schroeder wrote: > Eric Wilhelm wrote: >> BTW, the TAPx project would love to have a wxPerl gui test harness >> example. There's something that could be used by many perl >> programmers, even programmers of other languages, and has a small, but >> non-trivial issue of IPC with running children in a fork. IMO, it >> would also be a great place to apply a Wx::TreeListCtrl widget. >> > Oh -- we've had another contractor (Brad Embree) working the past few > months on a test harness for wxPerl. At this point it assumes you will > be using the Wx::Data controls (see other email) which are > "test-aware". You can "record" a test, which gets saved (currently > using YAML I think??). You can then "replay" a test and the test > harness will compare actual vs expected. I'll see if Brad can write up > something more meaningful about this. > > Mike. I am actually in the process of finishing the test harness up, and integrating it back into our main trunk for some final testing. The basic idea is that we have our own Event class, that exports a register_event function. We register all events using this function, instead of using the Wx event "macros". For example: EVT_BUTTON( $self, $self->{create_account_button}, \&OnCreateAccount ); becomes: register_event( $self, 'EVT_BUTTON', 'create_account_button', \&OnCreateAccount ); The register_event function then "wraps" the event handler inside another sub that records the detail of the event (right now just to a text file) before calling the handler. We can turn "recording" on and off dynamically as the app runs. We can select a text file for "playback", and it will run through all the entries in the text file replaying the recorded actions. What I am finishing up now is a comparison of the data that is returned from our API during playback to what was received during the test run (it is this API data that is stored as YAML). If anyone is interested I can put together a more thorough write up, and perhaps provide a simple example app demonstrating recording/playback. Brad |
From: Eric W. <scr...@gm...> - 2007-02-13 05:10:40
|
# from Foo JH # on Monday 12 February 2007 07:39 pm: >> Agreed. Actually, at this point, just about any solution looks >> better to me than POE. After walking the POE road, I am backing out >> now and re-working the problem with a combination of things. I >> think POE is pretty neat, but to turn my code on its head and still >> not get true concurrency is really just not turning out to be worth >> it. If I want to mangle my architecure, I better get more for it. > >Is there really an alternative to POE for some real threading in > wxPerl? Perl isn't famous for it's threading libraries, but a good > GUI should freeze when it's busy doing something else... POE isn't "real threading", it's cooperative multitasking. It's also a few other things, but the simplest usage doesn't involve parallelization, only CMT. If your code can be structured as iterators, you don't need threading and it is much simpler to say $thing->work in idle than to try to deal with real asynchronous processing. --Eric -- perl -e 'srand; print join(" ",sort({rand() < 0.5} qw(sometimes it is important to be consistent)));' --------------------------------------------------- http://scratchcomputing.com --------------------------------------------------- |
From: Foo JH <jhf...@ex...> - 2007-02-13 07:21:54
|
> If your code can be structured as iterators, you don't need threading > and it is much simpler to say $thing->work in idle than to try to deal > with real asynchronous processing. > Not really. In my client-side app which takes significant time to comms with a server, during this time if the client where to simply alt-tab to and fro, the wxperl app does not refresh, and it really looks like it hung. To enable the concept of cooperative multitasking, I'd prob have to break up the comms into smaller chunks, and have some form of state memory for the poe framework to work. This however is not exactly the straightforward style of coding as one would normally do (though this is prob an arguable pt). It seems to me that everyone knows how bad threading is for Perl, and rather than openly admit it, we regress (again, another arguable position) into cooperative multitasking. IMHO not the right thnig to do. > --Eric > |
From: Eric W. <scr...@gm...> - 2007-02-13 09:50:18
|
# from Foo JH # on Monday 12 February 2007 11:21 pm: >Not really. In my client-side app which takes significant time to > comms with a server, during this time if the client where to simply > alt-tab to and fro, the wxperl app does not refresh, and it really > looks like it hung. The connection can be asynchronous (at the socket level) without the Perl code needing to be threaded. >It seems to me that everyone knows how bad threading is for Perl, and >rather than openly admit it, we regress (again, another arguable >position) into cooperative multitasking. IMHO not the right thnig to > do. I don't think there's any denial in play. What do you propose? --Eric -- "Left to themselves, things tend to go from bad to worse." --Murphy's Corollary --------------------------------------------------- http://scratchcomputing.com --------------------------------------------------- |
From: Foo JH <jhf...@ex...> - 2007-02-13 12:30:53
|
> The connection can be asynchronous (at the socket level) without the > Perl code needing to be threaded. > You mean a non-blocking socket call. If I understand POE correctly you have to alter the flow of the code so that the application operates on the concept of states. So we're back to cooperative multitasking, which I find unnecessary given the maturity and stabilty of today's OSes, including Windows. >> It seems to me that everyone knows how bad threading is for Perl, and >> rather than openly admit it, we regress (again, another arguable >> position) into cooperative multitasking. IMHO not the right thnig to >> do. >> > > I don't think there's any denial in play. What do you propose? > I don't think with 5.8.x there's much that can be done. I do look forward to 5.10/ 6.0 where I expect threading to be on par (at least) with the current flavours of programming languages. Perl really needs to have something to show for this year, imho. Don't see me as a whiner on the downsides of Perl. There are quite a few workarounds on this issue, so life goes on. Even then, workarounds are merely short-term measures. |
From: Donovan A. <al...@Zo...> - 2007-02-13 20:55:58
|
We are sliding way sideways here... =20 Threads and cooperative multitasking are not mutually exclusive. To = really get optimal performance for your cpu cycles & memory buck, using = both is ideal. IMHO, this is more true when the CMT technology is also = an event loop. =20 However, I totally agree with you in principle Foo. Using rock spears = (and struggling to melt copper) in the iron age. If perl wasn't one = hell of a good rock spear this wouldn't be so much of a concern. =20 I must have missed something somewhere...is 5.10 & 6 going to have = something new on the threading front? =20 =20 ________________________________ From: wxp...@li... on behalf of Foo JH Sent: Tue 2/13/2007 4:30 AM To: Eric Wilhelm Cc: wxp...@li... Subject: Re: [wxperl-users] alternatives to POE > The connection can be asynchronous (at the socket level) without the > Perl code needing to be threaded. > =20 You mean a non-blocking socket call. If I understand POE correctly you have to alter the flow of the code so that the application operates on the concept of states. So we're back to cooperative multitasking, which I find unnecessary given the maturity and stabilty of today's OSes, including Windows. >> It seems to me that everyone knows how bad threading is for Perl, and >> rather than openly admit it, we regress (again, another arguable >> position) into cooperative multitasking. IMHO not the right thnig to >> do. >> =20 > > I don't think there's any denial in play. What do you propose? > =20 I don't think with 5.8.x there's much that can be done. I do look forward to 5.10/ 6.0 where I expect threading to be on par (at least) with the current flavours of programming languages. Perl really needs to have something to show for this year, imho. Don't see me as a whiner on the downsides of Perl. There are quite a few workarounds on this issue, so life goes on. Even then, workarounds are merely short-term measures. -------------------------------------------------------------------------= Using Tomcat but need to do more? Need to support web services, = security? Get stuff done quickly with pre-integrated technology to make your job = easier. Download IBM WebSphere Application Server v.1.0.1 based on Apache = Geronimo http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D= 121642 _______________________________________________ wxperl-users mailing list wxp...@li... https://lists.sourceforge.net/lists/listinfo/wxperl-users |
From: Donovan A. <al...@Zo...> - 2007-02-13 20:37:39
|
=20 > To enable the concept of cooperative multitasking, I'd prob have to > break up the comms into smaller chunks, and have some form of state > memory for the poe framework to work. This however is not exactly the > straightforward style of coding as one would normally do (though this = is > prob an arguable pt). POE is non-blocking operators + iterators + event loop =3D cooperative = multitasking. > It seems to me that everyone knows how bad threading is for Perl, and > rather than openly admit it, we regress (again, another arguable > position) into cooperative multitasking. IMHO not the right thnig to = do. I think threading in perl is not so bad, but it is still very = situational on where and when=20 you can really use it due to many modules that are not thread safe and = the extreme memory overhead if your threads don't need all of the cloned data. =20 Threads, no matter what language, have their down side. At least in = perl threads you have to be much more involved in mashing global data = (in general) because you have to share it. =20 I recently stumbled across CLONE_SKIP in threads, which may provide a = less bumpy ride=20 on these issues, but untested by me ATM. My current solution is a mixture of cooperative multitasking (event = loop) and parallelism (multiple peer apps using RPC over IPC-Shared = Memory, a message queue, some peer apps are threading). -------------------------------------------------------------------------= Using Tomcat but need to do more? Need to support web services, = security? Get stuff done quickly with pre-integrated technology to make your job = easier. Download IBM WebSphere Application Server v.1.0.1 based on Apache = Geronimo http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D= 121642 = <http://sel.as-us.falkag.net/sel?cmd=3Dlnk&kid=3D120709&bid=3D263057&dat=3D= 121642>=20 _______________________________________________ wxperl-users mailing list wxp...@li... https://lists.sourceforge.net/lists/listinfo/wxperl-users = <https://lists.sourceforge.net/lists/listinfo/wxperl-users>=20 |
From: Mattia B. <mat...@li...> - 2007-02-17 14:55:43
|
On Tue, 13 Feb 2007 12:37:15 -0800 "Donovan Allen" <al...@Zo...> wrote: > > > To enable the concept of cooperative multitasking, I'd prob have to > > break up the comms into smaller chunks, and have some form of state > > memory for the poe framework to work. This however is not exactly the > > straightforward style of coding as one would normally do (though this is > > prob an arguable pt). > > POE is non-blocking operators + iterators + event loop = cooperative multitasking. > > > It seems to me that everyone knows how bad threading is for Perl, and > > rather than openly admit it, we regress (again, another arguable > > position) into cooperative multitasking. IMHO not the right thnig to do. > > I think threading in perl is not so bad, but it is still very situational on where and when > you can really use it due to many modules that are not thread safe and the extreme memory > overhead if your threads don't need all of the cloned data. I tend to agree. I also wonder if anybody has looked at http://search.cpan.org/~mbarbon/Wx-App-Mastermind-0.02/ "a nontrivial example of wxPerl threads" Regards Mattia |
From: Huub P. <hu...@tr...> - 2007-02-13 23:32:38
|
Just a quick mail to let you all know, I haven't run off.... :-) I'm very happy to see so many responses with many good ideas, reasons to do or don't, arguments and what not. There was so little traffic on the list lately I was afraid I was allready to late. I've got a lot to ask, explain and respond to and I think I can in one or two days but since I initiated this thread I thought it was necessary to let you all know I'm still here... I think there's enough food for discussion and I think we've got something here to get started soon, in whatever shape or form or place it will turn out. Regards, Huub Peters |
From: Peter <lan...@fr...> - 2007-02-15 10:45:25
|
This reminds me a lot of the 'Sorry state of wxPerl' thread I started exactly a year ago (when the list wasn't so full of spam). I hope something comes out of this one. Regards, Peter On Wed, 14 Feb 2007 00:32:31 +0100, "Huub Peters" <hu...@tr...> said: > Just a quick mail to let you all know, I haven't run off.... :-) > I'm very happy to see so many responses with many good ideas, reasons to > do > or don't, arguments and what not. > There was so little traffic on the list lately I was afraid I was > allready > to late. > > I've got a lot to ask, explain and respond to and I think I can in one or > two days but since I initiated this thread I thought it was necessary to > let > you all know I'm still here... > > I think there's enough food for discussion and I think we've got > something > here to get started soon, in whatever shape or form or place it will turn > out. > > Regards, > > Huub Peters > > > > ------------------------------------------------------------------------- > Using Tomcat but need to do more? Need to support web services, security? > Get stuff done quickly with pre-integrated technology to make your job > easier. > Download IBM WebSphere Application Server v.1.0.1 based on Apache > Geronimo > http://sel.as-us.falkag.net/sel?cmd=lnk&kid=120709&bid=263057&dat=121642 > _______________________________________________ > wxperl-users mailing list > wxp...@li... > https://lists.sourceforge.net/lists/listinfo/wxperl-users |