tetragon, Infinoid: I've posted an updated version of tetragon's patch to RT 53086, please give it a try. G'night! % markus has joined #parrot % markus has left markus!~markus@80-123-36-91.adsl.highway.telekom.at % markus has joined #parrot r27055 | fperrad++ | trunk: : [opengl] : freeglut.h is not available on win32, only glut.h (which is checked by config/auto/opengl.pm) diff: http://www.parrotvm.org/svn/parrot/revision?rev=27055 r27056 | fperrad++ | trunk: : [perldoc] : - fix with strawberry Perl 5.10.0 diff: http://www.parrotvm.org/svn/parrot/revision?rev=27056 % DarkWolf84 has joined #parrot % pjcj has joined #parrot % rdice has joined #parrot % markus has left markus!~markus@80-123-36-91.adsl.highway.telekom.at % iblechbot has joined #parrot r27057 | allison++ | pdd25cx: : [pdd25cx] Bringing the pdd25cx branch up-to-date with trunk r27056. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27057 % askie has left askie!~askie@81.171.100.207 % askie has joined #parrot % rdice has left rdice!~richarddi@CPE001217e365c7-CM00159a01d44c.cpe.net.cable.rogers.com japhb: r27055 conflicts with your patch, but tested successfully with r27054 japhb: Determining if your platform supports OpenGL...............yes, freeglut 4. % rdice has joined #parrot % kid51 has joined #parrot % bgeron has left bgeron!bgeron@toad.stack.nl japhb: your RT#53086 patch causes test failures in t/steps/auto_opengl* and a couple of the codingstd tests. % bgeron has joined #parrot japhb: I think r27055 is discardable given your new $glut_header mechanism, but we're going to need to get fperrad to test it again. % DarkWolf84 has left DarkWolf84!~dwolf@89.215.234.147 r27058 | jkeenan++ | trunk: : The internals of this config step class are still in development. To prevent spurious test failures, am temporarily disabling tests of internal subroutines and methods. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27058 Infinoid: I just disabled the failing tests in t/steps/auto_opengl*. I'll rewrite them once the internals are more stable. thanks! I'll stop worrying about how to fix them, then As to this question: Does the config system have a mechanism for enumerating the (platform-specific) multitude of include-directories, to see which one contains a header needed for the opengl stuff? My answer is: Probably not yet (I think). I've never plunged that deeply into the config/gen/*.pm modules. Writing tests for them is still on my TODO list. I have developed and committed a way to determine which files are created during Configure.pl. But I don't think that's what you're getting at. no, we're talking about system include paths thanks. japhb really wants that, so he can find some opengl headers and parse the constants out of them. but I think he's currently planning to parse cflags and do it in a big for-loop, or something like that Ok, I've posted a fresh opengl patch to RT#53086. Everything works for me on linux/amd64. I am hoping it works on OSX and mingw32 too :) * tetragon wakes up I can tell you about OSX at some point in the next hour that'll be great, thanks! Infinoid: I just posted to 53086 as well. cool does examples/opengl/triangle.pir run for you? % wknight8111 has joined #parrot On which OS? With what patches applied? on OSX, with the patch you just posted Hmm. I just reverted, because everything I've been doing was in my main sandbox. I can see I'm going to have to have a separate sandbox for this stuff. oh, sorry And what exactly do you mean by examples/opengl/triangle.pir? [parrot] 505 $ ll examples/opengl/triangle.pir ls: examples/opengl/triangle.pir: No such file or directory [parrot] 506 $ ll t/examples/opengl/triangle.pir.t ls: t/examples/opengl/triangle.pir.t: No such file or directory infinoid@chirp parrot-trunk % ./parrot examples/opengl/triangle.pir How would that be run? (I've never poked around in examples/) as shown above just run parrot on it... it pops up an opengl window, here, and closes when you hit escape Well, when I get this sandbox, I'll have to apply patch, configure and build -- which on my iBook is a slow process. With kid51's patch: Determining if your platform supports OpenGL............yes, MacOSX_GLUT 5. kid51: sounds like tetragon is already testing it :) Yes, she should get similar results to me. Mine is just a slightly tidied version of japhb's. That's what I saw when I read it mine too... just rebased it and removed one line's worth of trailing whitespace * kid51 configures in new sandbox, having applied patch yours is the more useful patch. I notice that this step takes longer to run than others. Well, it does grep out some of the contents of all your OpenGL and GLUT headers * kid51 runs make really? the longest step for me is: Detecting supported compiler warnings (-Wxxx)..................set for gcc. but its hard to measure... Configure.pl completes in 6.5 seconds. I meant: longer than the other step classes in its vicinity: auto::readline, auto::crypto, etc. Infinoid has a fast box. * Infinoid hugs his laptop My iBook is fast You should see my other laptops Mine is slow. My Linux VM is fast. * tetragon likes her backup laptop, a Pentium 100 hooray for work-issued thinkpads But for really fast, I'd have to use my G5/Xeon PowerMac cluster at work * kid51 afk ooh. they're going to come out with a quad core thinkpad I want one. it'll burn a hole in my leg even faster than this one does I just want a durable macbook, my current macbook platic chips and in no more blank. s/in/is/ Yay, spinning triangle Apparently the next generation will look more like an airbook. No more cheap plastic. great, thanks! btw, is ** implemented in PGE ?? * Infinoid is PGE-clueless, sorry now as soon as we hear back from fperrad, we can think about checking this stuff in If I were to replace this one, I'd propably go with the macbook pro. Nicer keyboard, and the graphics chip is better suited for what I do Apparently there is **{m..n}, but not ** ',' . I am interested in the later. the later is soooooo cool kid51: (RT #53094) one minor addition to what I said on-list: fperrad's change will break if running on a machine that doesn't have any perls in its $PATH. (I don't know if such a machine exists, though.) ./parrot examples/opengl/triangle.pir works for me: spinning triangle. But it launches very slowly. % jan has left jan!~chatzilla@90-229-128-103-no52.tbcn.telia.com I didn't notice a terribly slow launch Well, we already know your iBook is faster than mine! How old is your iBook? % Limbic_Region has joined #parrot 4 years Mine is about 2.5 years you have powerpc? Yes you bought just before the move to intel. I know My ppc mac works better than my intel mac I bought just after and I had many problems. Next time, I will wait two months. And this iBook case has no chips out of it, just some scuffing and scratches in november, my mac dropped to the shell once the week. I hesited to invest to 10.5 but apparently it updated the firmware and fixed that problem so I am an enthousiat for the software not for the hardware I updated my iBook to 10.5 back in March, my iMac still hasn't been updated not that I explored much the proprietary software. anyway, I am a X hater. so that was good riddance. hate X11 or hte OS X? X11 stole the precious, we hates it (precious=200M of my RAM) % DarkWolf84 has joined #parrot % nopaste has left nopaste!~opaste@202-154-105-237.people.net.au % TonyC has left TonyC!~tony@202-154-105-237.people.net.au % nopaste has joined #parrot % kid51 has left kid51!~jkeen@pool-70-107-13-215.ny325.east.verizon.net this box doesn't have a patch command. how do people on win32 apply patches? is there something on CPAN I can install? google "patch for windows" % kid51 has joined #parrot * kid51 hates when his crummy Verizon-supplied wireless router drops the Internet connection. Tene, I was marginally involved in NeWs the ill-fated Sun Window system : http://en.wikipedia.org/wiki/NeWS if I had an extra couple of hours, I'd write an App::Patch wrapper around Text::Patch, so that I wouldn't have to fight against this gnuwin32-patch restriction about crlf line termination. X was adopted as a reaction to NeWS and Sun hubris when in bed with ATT cooperative multitasking postscript interpreter? wow. when are we getting a languages/postscript/ ? :) Sun made many mistake like not paying a wad of cash to adobe to get a decent scalable font system. the interesting part of postscript is not the language but the graphic model. % kid51 has left kid51!~jkeen@pool-70-107-13-215.ny325.east.verizon.net NeWs had me for me an interesting life case of how to transform a brillant idee in a crummy product marketted in an inept way. s/idee/idea/ the event-heirarchy blurb on the wikipedia page makes me think of Berlin % wknight8111 has left wknight8111!~nobody@c-71-230-33-251.hsd1.pa.comcast.net or Fresco, or whatever they're calling it now I think Berlin stole a few idea from NeWS, the main one being to never get market share :) seems like: $Fresco = $NeWS =~ s/postscript/corba/ ooch, I never liked corba being burned by xml/rpc. s/=~/~~/ btw :) ooh, recursive regexes % kid51 has joined #parrot some university did a infixed syntax layer on top of the PostScript of NeWS, but Sun never picked-it up. you must have grown with a HP25 to accept PostScript as a programming language for human versus being generated by programs. syntax matters, in fact, most people picks one language not for semantic but by likin its syntax. * cognominal wants ** ',' and looks how hard it is to implement it in PGE heh I never learned postscript for the same reasons I never learned corba: big and complicated, and I haven't found time for it yet The syntax that I end up dealing with but don't like is applescript anaway, now this PostScript is supplanted by Pdf which is even bigger tetragon : if perl is write-only, applescript is definitevely read-only it is difficult to figure what subset of english iti is. http://search.cpan.org/dist/Fuse-PDF/ is one of the more bizarre things I've heard of apparently .iso wasn't good enough for him :) may be pdf is turing complete. cognominal: applescript is one of the set of languages I use at work and with enough input output primitive. Guess, test, slam head against the stack of dictionaries I heard there is some good scripting language on Apple. Don't remember the name. I use bash, perl, python, and applescript heh hope you have a padded dictionary But with 10.5 and some work, I can do everything that I can with applescript in python. More work would allow for Perl % jan has joined #parrot rubycocoa seems nice too because camelbones was lagging last time I checked I use pyobjc I can't use camelbones due to its current status btw, I have studied dtrace which is on Leopard. I should work harder to add it to parrot. Just don't run iTunes while testing Instrument breaks the ones playing on iTunes? cognominal: feel free to file a ticket for ** in PGE http://blogs.sun.com/ahl/entry/mac_os_x_and_the I think there are workarounds for that now pmichaud, I follow a pattern to patch each time a very different part of parrot. So why not PGE :) eventually I will get a good idea of what is parrot. pmichaud : and your code is well organied and commented. % markus has joined #parrot cognominal: I think you're the first person to claim that my code is well commented :-) but thanks :-) it had to be, because you deal with hairt stuff hairy! whenever I go back and look at code I tend to add comments to it and I am immunized, I have seen perl5 interpretor code with tolkien quotes as sole comments. opengl wasn't detected on mingw. Maybe fperrad has some extra libraries installed for that hopefully nicholas, rgs and many other have fixed that. cognominal: that's a very slow process, I think Infinoid: what is a slow process? properly commenting an existing project like p5 I bet % Andy has joined #parrot thx pmichaud for commenting my rt tickets * japhb now done reading IRC backlog, now for email ... japhb: please consider jkeen@verizon.net's patch, mine is out of date. the only area I'm not sure about is whether we need that #ifdef WIN32... I think your config::gen code should be able to handle that without the need for a special case. and good morning, by the way. :) Good morning, and yes, I'm working through the RT emails now. :-) % rdice_ has joined #parrot And that's correct, I think we can avoid #ifdef WIN32; I'm trying everywhere I can to git rid of OS #ifdefs, and have only #ifdefs linked to identifiers created by the GLUT headers themselves. % markus_ has joined #parrot the #ifdef snuck in there between the time when you posted the patch and when we started testing it I'm hoping to hear back from fperrad on whether its actually necessary. but if it looks good to you, I think the rest of it can be checked in right now, and we can remove the ifdef later % markus has left markus!~markus@80-123-36-91.adsl.highway.telekom.at % ambs has joined #parrot gah, dang puking cat hmmms... my MacOS continues not detecting OpenGL probably I do not have it :-S % ambs has left ambs!~ambs@255.92.54.77.rev.vodafone.pt Gah, was about to respond me too. % ambs has joined #parrot Ah, there you are ambs, have you installed the OpenGL and GLUT frameworks? japhb, that is my doubt where should it be? That's a question for one of the others. :-) :D ambs: OSX won't work without the patch attached to RT #53086, so please apply that if you don't already have it oh ambs: the *latest* patch cognominal: (rt tickets) -- glad to comment, but as you can see the ticket you submitted depends on things that have been stalled for months I don't know when there will be progress on that. afk, lunch and errands and probably soccer. I am happy to submit to patch but also to be as nasty as an end user. * ambs patches Infinoid: OK, reviewed the latest patch. The #ifdef WIN32 should go away. We need to make sure that the $glut_header detection Just Works everywhere. YAY! Success Determining if your platform supports OpenGL............yes, MacOSX_GLUT 5. rawk Infinoid: can you regenerate the patch without the #ifdef WIN32? I'm juggling some Real Life stuff this morning, so it will take me a bit, but I can do it if you don't have the cycles can do. and then I'll check that in ambs: great! the ultimate test is whether "./parrot examples/opengl/triangle.pir" works. Infinoid: thanks! Infinoid, still compiling ambs: thanks. I think it should work, we have 2 success reports (kid51 and tetragon) for OS X so far but more testing never hurts. :) NODNOD Infinoid, it is not properly a 3D triangle, but it is a trinangle, indeed :-D japhb: new bug report: make your triangle fancier! :) ambs: It is actually a 3D triangle ... I just hadn't written the code to turn on perspective. :-) * Infinoid removes #ifdef WIN32, and some trailing whitespace How about a sierpinski triangle? Infinoid: heh hehehe tetragon: I think freeglut includes code to make a sierpinski triangle and sponge. :-) next step: translate quake to .pir jajaja now, should I commit the patch? :D * ambs grins Infinoid: In the perl 5 world, I was partway through porting Classic Doom to Perl 5, so hey, that's part of the battle. ambs: I'm working on that, we have some other minor modifications but nothing that'll affect OSX * Infinoid is just waiting for "make test" to complete Infinoid, sure :) actually, quake would make a great benchmarking tool Infinoid: yes, it definitely would. And I have similar aspirations. :-) japhb: great! so you'll have it ported by this time tomorrow, right? (j/k) :D Odd factoid (my by an urban legend, but I have no particular reason to doubt it) -- apparently in the 486 timeframe, Classic Doom exercised more bits of the chip simultaneously and harder than any other app out there -- so Intel used it as a burn-in test for 486 chips. If it could survive looping Doom for a while, it probably wouldn't fail running anything else. :-) Infinoid: heh. Not quite. :-) Er ... "may be an urban legend" interesting. I read somewhere that quake and related engines ran in a virtual machine. so it might be better suited for parrot than I had originally thought :) Different versions of quake did it differently. Some used VMs, some just had native code in DLLs. stop talking about quake, or I might get out to play quake 3 arena :) Personally, I think we should add QuakeC to the list of Parrot languages. :-) The one that's really important is z-machine I thought someone was already working on that ...? languages/Zcode/ Doesn't look done yet r27059 | infinoid++ | trunk: : [OpenGL] Apply patch from japhb++ (RT#53086) with the following modifications: : * kid51++ contributed some changes to configure more cleanly : * remove some trailing whitespace : * revert fperrad++'s r27055, config::gen::opengl should be smart enough to handle that internally : tested by myself on linux/amd64; tested by tetragon++ kid51++ and ambs++ on OS X. diff: http://www.parrotvm.org/svn/parrot/revision?rev=27059 Thank you, all of above. erm, conclicts ambs, yes, because you had a previous rev of the patch installed. :-) ok done yeah Infinoid++ for commiting it :-P I already got one of those :P I know and still working on linux, without display. nice awesome :-P w00t so, just to drive ambs crazy, here's another quake-related note I'm working on PDD13, we want to revamp the .pbc file handling stuff, and put that code into .pmc classes we could probably do the same thing for .wad files, pretty easily .wad is Doom. .pak and .pk3 are Quake oh yeah But point remains the same. :-) brb (eat something) .wad is a binary format with unique task-specific compression. The quake formats where just ZIP, more or less. yeah, I've played with both at distant points in the past, and my brain is mixing them up my bad :) np -- If you've worked with them, I will probably want to pick your brains about some edge cases in the future. :-) I doubt I'd be much help Last time I worked on the Doom stuff, I was still trying to write code to create crack-free floors/ceilings ... original Doom .wads made use of quirks of id's renderer to avoid having have perfectly complete data. (the id renderer sortof "filled in the cracks" at render time) * kid51 wakes up and tests opengl now that patches have been committed to trunk % Andy has left Andy!~AndyL@host3130.follett.com * ambs is back * ambs wonders if the University cluster has opengl libraries % jq- has joined #parrot % jq- has left jq-!~jquelin@merlin.mongueurs.net * Infinoid wonders how fperrad set up glut on strawberry perl, because it isn't there by default Mind you, for the last couple of days, I've just been working on getting the basic configure/build structure working. There's only enough OpenGL bound to run the example. Now that we seem to be closing in on that, I'll go back to working on the main bindings. With thanks to all of you .... Determining if your platform supports OpenGL...............yes, freeglut 4. :-O Determining if your platform supports OpenGL............yes, MacOSX_GLUT 5. ... so far so good Determining if your platform supports OpenGL............yes, MacOSX_GLUT 5. now, and regarding OpenGL: we really want to link parrot with the openGL library? Shouldn't we create just a binding library for it? (sorry if it is being done that way already) ambs: OpenGL.pir assumes the binding is dynamically loaded. But not being an OS X guy, I have no idea what's really going on under the covers there. It Seems To Work, but who knows? Whatever the 'loadlib' op does on OS X. oh, probably it is being done in the way, then. in the *correct way I would *not* put it past the darwin folks to statically link and then make dynamic loading ops magically work. They're bloody strange like that. ;-) japhb: The test program used by Configure.pl builds, links, and runs just fine without any flags tetragon: what do you mean by "without any flags"? No warnings? As in "gcc -o test foo.c" No warnings tetragon: that's ... odd. % muixirt has joined #parrot otool -L shows that the output is not linked (against the frameworks) hi But it runs just fine tetragon: I bow to the much more knowledgeable of us two. :-) So does this mean we could try *not* doing -framework OpenGL -framework GLUT in auto::opengl? Can that whole darwin check go away? perl6 -e 'say 1*7.2' gives me 9.96345e+08 muixirt: Didn't you learn that in elementary school? ;-) :-) japhb: You still need some darwin checks. Things like -lgl don't work so what is the explanation? tetragon: fair enough, but what happens when darwin adds no linker libs at all? (i.e. we check for it, and in the darwin case do nothing) perl6 -e 'say 7.2*12345679' 8.8889e+07 perl6 -e 'say 12345679*7.2' 9.96345e+08 hmm japhb: If I tell config/auto/opengl.pm to add no flags, Configure.pl still detects OpenGL tetragon: nod, but can you get all the way to running triangle.pir make is still running nodnod And it has just finished building glut_callbacks.o I'm getting the spinning triangle. This is interesting, because it's the first time that in my Mac GUI a program called 'parrot' is running. This program called 'parrot' responds to Cmd-, (i.e., it has a Preferences window). And it also responds to Cmd-Q. cool! In the meantime, I'm still running 'make test' to see if everything else is still working. The triangle still spins tetragon: ooh, that's interesting muixirt: looks like "say" isn't handling floating point properly. or else, arithmetic itself isn't handling floating point properly. what's your platform? japhb: That OS X does a lot of dynamic lookups? tetragon: hold on for a sec, phone call Infinoid, GNU/Linux with Rakudo Perl 6, revision 27058 built on parrot 0.6.1-devel % Zaba has joined #parrot moin muixirt: linux/x86? Infinoid, yes ok. I'm seeing similar behavior on linux/amd64. and the fun part is, the output is random infinoid@chirp perl6 % for SEQ in 1 2 3 4 5; do ../../parrot perl6.pbc -e 'say 1*2.1'; done 9.86553e+13 1.0056e+14 9.98779e+13 1.00221e+14 9.85657e+13 1.1*1 works fine. 1*1.1 fails horribly. smells like something uninitalized ... I am running SVN parrot on x86_64 (compiled with gcc 4.1.2), it passed the test suite. I compiled languages/perl6, and when I do 'sub test {}', then 'test();', and then ctrl+d on interactive prompt, parrot aborts with the following backtrace: http://rafb.net/p/2A5tfk82.html it does not happen on just ctrl+d isn't random for me, but the value depends on the float value Hrm... "make perl6" seems to be either hanging or being very slow is it a known issue? tetragon: compiling rakudo is very slow muixirt: does 1.1*1 print normally? is it just 1*1.1? 1.1*1 is ok, 1*1.1 is excessibely big Zaba: Can you paste what you did leading up to that abort? Infinoid, yes * seems to give right answers but not * Infinoid: It was just slow kid51, http://rafb.net/p/4nsuRc41.html on ./perl6 interactive prompt Well, I'm actually unfamiliar with the interactive prompt. So I can't give you a good answer. If there's no one else on channel now who can answer, I'd suggest opening a bug ticket: email to parrotbug@parrotcode.org with [BUG] in subject line. Attach your pastes in an email attachment, preferably with name ending in '.txt'. OK, off call. tetragon, you were saying? Zaba: I tried the sub test {} test() thing I got a load of double free and non-aligned pointer errors tetragon, I get that too muixirt: kid51's advice is what I would do for your bug, too :) though I find gdb's bt more usable * tetragon goes and sets the malloc debugging environment variables kid51, having those lines in a file causes it to crash with the same reason that is, sub test {}\ntest(); tetragon: my question with the opengl thing on OS X is ... does compiling without -framework OpenGL -framework GLUT reduce the size of parrot or libparrot? Since you said it still runs .... japhb: I don't have the linked one's size, but libparrot is 6MB and parrot is 29kb kid51, so it's not really an interactive mode-specific problem tetragon: when you get a chance, that would be very interesting info to me. Zaba: Okay, so please open an RT. Ok. But either way, you might want to commit without the -framework flags and see if it still works for everyone. I always prefer less stuff on the link lines -- link lines should be truthful about what is really being added. Much about ./perl6 is unexplored territory, because it's only been since late December that we've had a way for people to easily get to "Hello world." So RT reports of unexpected behavior are very welcome. * Zaba sees a befunge implementation in languages/ note: muixirt's bug is reproducible with ../../parrot perl6.pbc, as well as ./perl6 Zaba: you can blame jq for that. but it's pretty out of date So my first malloc error is a non-aligned pointer being freed Infinoid, I wonder what specs it implements :) % skids has left skids!~bri@c-71-233-204-100.hsd1.ma.comcast.net % skids has joined #parrot Befunge-93 oh, I see. A friend of mine is writing/maintaining a Befunge-98 interpreter in C only befunge? (I am mainly interested in trefunge+) Zaba: Confirmed your results on Linux i386. $ ./perl6 emptysub *** glibc detected *** corrupted double-linked list: 0x08563b48 *** Aborted kid51, it does happen when a sub contains 'say "moep"' as well so not only empty.. I'm looking at it with some added memory debugging and in gdb When I reverse the order of lines inside the file 'emptysub', I get: It's right when perl6 is exitting $ ./perl6 emptysub Segmentation fault (Core dumped) uh oh.. perl6(96032) malloc: *** error for object 0x1791500: Non-aligned pointer being freed (2) Zaba: I do some AI work with a 4-dimensional funge space, using Language::Befunge That object is interp Infinoid, the befunge-98 interpreter I mentioned is very fast Infinoid, its main aim is speed and standard compiliance sounds useful. but can it handle more than 2 dimensions?