Skip Menu |

Preferred bug tracker

Please visit the preferred bug tracker to report your issue.

This queue is for tickets about the PDL CPAN distribution.

Report information
The Basics
Id: 30353
Status: resolved
Priority: 0/
Queue: PDL

People
Owner: chm [...] cpan.org
Requestors: twc [...] aaocbn.aao.gov.au
Cc:
AdminCc:

Bug Information
Severity: Normal
Broken in: 2.4.3
Fixed in: (no value)



Subject: WITH_3D libGL tests
Date: Tue, 30 Oct 2007 14:15:57 +1100 (EST)
To: bug-PDL [...] rt.cpan.org
From: Tim Connors <twc [...] aaocbn.aao.gov.au>
in Graphics/Makefile.PL, guessogllibs() tries to work out which libs to compile with to get -lGL. In my case, testfile appears to need: gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL -lXext -lX11 -lpthreads (this is with the nvidia GL libraries in /usr/lib/libGL) pthread is only tested for in the mesa case, which we don't have here. But furthermore, as far as I can work out, pthread would never be included anyway, because in libs(), ExtUtils::Liblist->ext("...-lpthread -lGL...") finds the libpthread, but then works out it is already within perl, and so doesn't include it in the output: /usr/lib/perl5/5.8.8/ExtUtils/Liblist/Kid.pm 170 # Do not add it into the list if it is already linked in 171 # with the main perl executable. 172 # We have to special-case the NeXT, because math and ndbm 173 # are both in libsys_s 174 unless ($in_perl || .... This then means that the standalone testfile.c is not linked in with pthread. Anyway, I added a libGL with pthread test, and this is what I get, after ExtUtils::Liblist->ext strips out the pthread library from the linker arguments -- Trying libGL with pthread... gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL -lXext -lX11 ... /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function `_glthread_InitTSD':: undefined reference to `pthread_key_create' /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function `_glthread_GetTSD':: undefined reference to `pthread_getspecific' /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function `_glthread_SetTSD':: undefined reference to `pthread_setspecific' collect2: ld returned 1 exit status NO So, what is being done wrong here -- is MakeMaker being used incorrectly? -- Tim Connors | Anglo-Australian Observatory http://site.aao.gov.au/twc | Coonabarabran, NSW 2357, Australia | Tel: +61 2 6842 6286
RT-Send-CC: pdl-porters [...] jach.hawaii.edu
On Mon Oct 29 23:16:30 2007, twc@aaocbn.aao.gov.au wrote: Show quoted text
> > in Graphics/Makefile.PL, guessogllibs() tries to work out which libs > to compile with to get -lGL. > > In my case, testfile appears to need: > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL
-lXext -lX11 -lpthreads Show quoted text
> (this is with the nvidia GL libraries in /usr/lib/libGL) > > pthread is only tested for in the mesa case, which we don't > have here. But furthermore, as far as I can work out, > pthread would never be included anyway, because in libs(), > ExtUtils::Liblist->ext("...-lpthread -lGL...") finds the libpthread, > but then works out it is already within perl, and so doesn't include > it in the output: > > /usr/lib/perl5/5.8.8/ExtUtils/Liblist/Kid.pm > 170 # Do not add it into the list if it is already linked in > 171 # with the main perl executable. > 172 # We have to special-case the NeXT, because math and ndbm > 173 # are both in libsys_s > 174 unless ($in_perl || .... > > This then means that the standalone testfile.c is not linked in with > pthread. Anyway, I added a libGL with pthread test, and this is what > I get, after ExtUtils::Liblist->ext strips out the pthread library > from the linker arguments -- > > Trying libGL with pthread... > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL > -lXext -lX11 ... > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function > `_glthread_InitTSD':: undefined reference to `pthread_key_create' > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function > `_glthread_GetTSD':: undefined reference to `pthread_getspecific' > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function > `_glthread_SetTSD':: undefined reference to `pthread_setspecific' > collect2: ld returned 1 exit status > NO > > So, what is being done wrong here -- is MakeMaker being used > incorrectly?
Thanks again for another problem report. I did a quick google for Nvidia Linux information and apparently the Nvidia drivers are a replacement for the MesaGL drivers so the Mesa flag options are probably correct. Note that the autoprobing can be skipped by hand editing the perldl.conf file in the top level PDL directory and setting WITH_3D to true (1), and manually defining the OPENGL_LIBS, OPENGL_DEFINE, and OPENGL_INC as in the examples given. If you hardwire in the Mesa stuff (modified for your system paths as needed) does the TriD module build successfully? As for the libs function, it won't pick and choose libraries from the test strings given so when guessogllibs tries '-lMesaGLU -lMesaGL -lpthread' and it works, the -lpthread is not omitted. Hope this helps and let me know what strings you were able to get things working with. Regards, Chris Marshall
CC: pdl-porters [...] jach.hawaii.edu
Subject: Re: [Pdl-porters] [rt.cpan.org #30353] WITH_3D libGL tests
Date: Tue, 6 Nov 2007 12:15:06 +1100 (EST)
To: Chris Marshall via RT <bug-PDL [...] rt.cpan.org>
From: Tim Connors <tconnors [...] astro.swin.edu.au>
On Mon, 5 Nov 2007, Chris Marshall via RT wrote: Show quoted text
> > <URL: http://rt.cpan.org/Ticket/Display.html?id=30353 > > > On Mon Oct 29 23:16:30 2007, twc@aaocbn.aao.gov.au wrote:
> > > > in Graphics/Makefile.PL, guessogllibs() tries to work out which libs > > to compile with to get -lGL. > > > > In my case, testfile appears to need: > > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL
> -lXext -lX11 -lpthreads
> > (this is with the nvidia GL libraries in /usr/lib/libGL) > > > > pthread is only tested for in the mesa case, which we don't > > have here. But furthermore, as far as I can work out, > > pthread would never be included anyway, because in libs(), > > ExtUtils::Liblist->ext("...-lpthread -lGL...") finds the libpthread, > > but then works out it is already within perl, and so doesn't include > > it in the output: > > > > /usr/lib/perl5/5.8.8/ExtUtils/Liblist/Kid.pm > > 170 # Do not add it into the list if it is already linked in > > 171 # with the main perl executable. > > 172 # We have to special-case the NeXT, because math and ndbm > > 173 # are both in libsys_s > > 174 unless ($in_perl || .... > > > > This then means that the standalone testfile.c is not linked in with > > pthread. Anyway, I added a libGL with pthread test, and this is what > > I get, after ExtUtils::Liblist->ext strips out the pthread library > > from the linker arguments -- > > > > Trying libGL with pthread... > > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -lGL > > -lXext -lX11 ... > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function > > `_glthread_InitTSD':: undefined reference to `pthread_key_create' > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function > > `_glthread_GetTSD':: undefined reference to `pthread_getspecific' > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function > > `_glthread_SetTSD':: undefined reference to `pthread_setspecific' > > collect2: ld returned 1 exit status > > NO > > > > So, what is being done wrong here -- is MakeMaker being used > > incorrectly?
> > Thanks again for another problem report. I did a quick google for > Nvidia Linux information and apparently the Nvidia drivers are a > replacement for the MesaGL drivers so the Mesa flag options are > probably correct.
Well, All the GL libs have the same API -- you install Mesa if you don't have any of other libs. At linking time, you appear to need -lGLU -lGL (-lpthread) over the top of any other X libs, where mesa wants -lMesaGLU -lMesaGL (-lpthread). Show quoted text
> Note that the autoprobing can be skipped by hand editing the > perldl.conf file in the top level PDL directory and setting WITH_3D > to true (1), and manually defining the OPENGL_LIBS, OPENGL_DEFINE, > and OPENGL_INC as in the examples given. If you hardwire in the > Mesa stuff (modified for your system paths as needed) does the TriD > module build successfully?
There's an unfortunate test in /Makefile.PL: for(keys %PDL_CONFIG) { if(!exists($PDL_CONFIG_DIST{$_})) { die << "EOD"; Invalid key $_ found in user supplied $pdl_conf_file - this key appears to be no longer in use. Please review configuration options and check the comments in the file perldl.conf that came with this distribution EOD } } The supplied ./perldl.conf file doesn't have OPENGL_LIBS declared (perhaps the "Let perl try to figure it out" *=undef lines should be uncommented in the supplied ./perldl.conf file), and so when you provide one in ~/.perldl.conf, it bails with that error message. This makes it impossible to test via the cpan shell, as far as I can tell, because I can't work out how to stop cpan from unpacking the .tar file everytime. I tried modifying the .tar file it has cached, but of course, it checks its md5sum. Anyone know how to override cpan? When it comes time to rolling this out, I'll be wanting to do it via cpan, but for testing, I'll show the results from a manual perl Makefile.PL below... Show quoted text
> As for the libs function, it won't pick and choose libraries from > the test strings given so when guessogllibs tries '-lMesaGLU -lMesaGL > -lpthread' > and it works, the -lpthread is not omitted.
It is though in the ExtUtils::Liblist->ext("...-lpthread -lGL...") test, which returns a string with the -lpthread removed, which then gets passed to gcc, which then links in the test program without libpthread, and fails, so the test comes back negative, and Makefile.PL thinks those supplied libraries don't work. So the result is: WITH_3D: Looking for xdpyinfo to check GLX WITH_3D: Trying OpenGL initial OPENGL_LIB configuration ->-L/usr/X11R6/lib -lGL -lXext -lX11 -lpthread -lm<- processed LIBS: -L/usr/X11R6/lib -L/usr/X11R6/lib -lGL -lXext -lX11 -lXext -lX11 Trying -L/usr/X11R6/lib -lGL -lXext -lX11 -lpthread -lm... gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib -L/usr/X11R6/lib -lGL -lXext -lX11 -lXext -lX11 ... /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function `_glthread_InitTSD':: undefined reference to `pthread_key_create' /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function `_glthread_GetTSD':: undefined reference to `pthread_getspecific' /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function `_glthread_SetTSD':: undefined reference to `pthread_setspecific' collect2: ld returned 1 exit status NO User specified OpenGL configuration did not work trying to guess... Darn, still trying to outsmart me. See how the -lpthread is not passed through to gcc? -- Tim Connors
On Mon Nov 05 20:15:49 2007, twc@aaocbn.aao.gov.au wrote: Show quoted text
> On Mon, 5 Nov 2007, Chris Marshall via RT wrote: >
Show quoted text
> > Note that the autoprobing can be skipped by hand editing the > > perldl.conf file in the top level PDL directory and setting WITH_3D > > to true (1), and manually defining the OPENGL_LIBS, OPENGL_DEFINE, > > and OPENGL_INC as in the examples given. If you hardwire in the > > Mesa stuff (modified for your system paths as needed) does the TriD > > module build successfully?
> > There's an unfortunate test in /Makefile.PL: > > for(keys %PDL_CONFIG) { > if(!exists($PDL_CONFIG_DIST{$_})) { > ...
Show quoted text
> This makes it impossible to test via the cpan shell, as far as I can > tell, > because I can't work out how to stop cpan from unpacking the .tar file > everytime. I tried modifying the .tar file it has cached, but of > course, > it checks its md5sum. Anyone know how to override cpan?
I think you can use the separate commands in the cpan shell to do this: 1. get PDL 2. look PDL (this will spawn a shell in the PDL build dir) 3. edit the perldl.conf as desired 4. Ctrl-D to exit shell 5. make PDL 6. test PDL 7. install PDL Show quoted text
> When it comes time to rolling this out, I'll be wanting to do it via > cpan, but for testing, I'll show the results from a manual perl > Makefile.PL below...
I would like to resolve this before the 2.4.4 point release. Thanks for the work to track these issues down. You've been a gold mine so far. :-) Show quoted text
>
> > As for the libs function, it won't pick and choose > > libraries from the test strings given so when guessogllibs > > tries '-lMesaGLU -lMesaGL -lpthread' > > and it works, the -lpthread is not omitted.
> ...
Show quoted text
> So the result is: > > WITH_3D: Looking for xdpyinfo to check GLX > WITH_3D: Trying OpenGL initial OPENGL_LIB configuration > ->-L/usr/X11R6/lib -lGL -lXext -lX11 -lpthread -lm<- > > processed LIBS: -L/usr/X11R6/lib -L/usr/X11R6/lib -lGL -lXext -lX11 > -lXext -lX11 > Trying -L/usr/X11R6/lib -lGL -lXext -lX11 -lpthread -lm... > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib > -L/usr/X11R6/lib -lGL -lXext -lX11 -lXext -lX11 ... > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function > `_glthread_InitTSD':: undefined reference to `pthread_key_create' > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function > `_glthread_GetTSD':: undefined reference to `pthread_getspecific' > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function > `_glthread_SetTSD':: undefined reference to `pthread_setspecific' > collect2: ld returned 1 exit status > NO > User specified OpenGL configuration did not work > trying to guess... > > > Darn, still trying to outsmart me. See how the -lpthread is not > passed through to gcc?
Hmm, I think you are right about the miracle library stripping powers of libs(). I suggest hand editing $libs at the end of guessogllibs() and add on -lpthread as was done for Linux and Solaris. The tests here need to be revisited before 2.4.4 but with hardwired flags in perldl.conf and this hack to guessogllibs() you should be able to move forward. Would you be willing to work from the CVS version of PDL instead of the current CPAN one? It would simplify getting fixes in for you and their migration into a 2.4.4 release by the end of this year. Regards, Chris
Subject: Re: [rt.cpan.org #30353] WITH_3D libGL tests
Date: Mon, 12 Nov 2007 03:56:52 +1100 (EST)
To: Chris Marshall via RT <bug-PDL [...] rt.cpan.org>
From: Tim Connors <twc [...] aaocbn.aao.gov.au>
On Tue, 6 Nov 2007, Chris Marshall via RT wrote: Show quoted text
> > <URL: http://rt.cpan.org/Ticket/Display.html?id=30353 > > > On Mon Nov 05 20:15:49 2007, twc@aaocbn.aao.gov.au wrote:
> > On Mon, 5 Nov 2007, Chris Marshall via RT wrote: > >
>
> > > Note that the autoprobing can be skipped by hand editing the > > > perldl.conf file in the top level PDL directory and setting WITH_3D > > > to true (1), and manually defining the OPENGL_LIBS, OPENGL_DEFINE, > > > and OPENGL_INC as in the examples given. If you hardwire in the > > > Mesa stuff (modified for your system paths as needed) does the TriD > > > module build successfully?
> > > > There's an unfortunate test in /Makefile.PL: > > > > for(keys %PDL_CONFIG) { > > if(!exists($PDL_CONFIG_DIST{$_})) { > > ...
>
> > This makes it impossible to test via the cpan shell, as far as I can > > tell, > > because I can't work out how to stop cpan from unpacking the .tar file > > everytime. I tried modifying the .tar file it has cached, but of > > course, > > it checks its md5sum. Anyone know how to override cpan?
> > I think you can use the separate commands in the cpan shell > to do this: > 1. get PDL > 2. look PDL (this will spawn a shell in the PDL build dir) > 3. edit the perldl.conf as desired > 4. Ctrl-D to exit shell > 5. make PDL
Ah, OK, that works (I don't think it was working when I was just doing make PDL - I seem to recall it was initiating a get as well -- perhaps the look PDL changes that, or perhaps doing an explicit get) ... Show quoted text
> > Trying -L/usr/X11R6/lib -lGL -lXext -lX11 -lpthread -lm... > > gcc -o /tmp/testfile /tmp/testfile.c -L/usr/X11R6/lib > > -L/usr/X11R6/lib -lGL -lXext -lX11 -lXext -lX11 ... > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x1f): In function > > `_glthread_InitTSD':: undefined reference to `pthread_key_create' > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0x6d): In function > > `_glthread_GetTSD':: undefined reference to `pthread_getspecific' > > /usr/X11R6/lib/libGL.a(glthread.o)(.text+0xa5): In function > > `_glthread_SetTSD':: undefined reference to `pthread_setspecific' > > collect2: ld returned 1 exit status > > NO > > User specified OpenGL configuration did not work > > trying to guess... > > > > > > Darn, still trying to outsmart me. See how the -lpthread is not > > passed through to gcc?
> > Hmm, I think you are right about the miracle library stripping > powers of libs(). I suggest hand editing $libs at the end of > guessogllibs() and add on -lpthread as was done for Linux and > Solaris. The tests here need to be revisited before 2.4.4 but > with hardwired flags in perldl.conf and this hack to guessogllibs() > you should be able to move forward.
Seems to be working. Compiling now... Show quoted text
> Would you be willing to work from the CVS version of PDL > instead of the current CPAN one? It would simplify getting > fixes in for you and their migration into a 2.4.4 release by > the end of this year.
Can do. http://pdl.sourceforge.net/FAQ has login instructions to cvs.pdl.sourceforge.net, but it looks like sourceforge has closed pserver access, so that could be out of date. BTW, that faq has a bad link in the text 3.3 ``CVS availability of PDL'' - should point to http://pdl.sourceforge.net/FAQ/#q__4_8_can_i_get_pdl_via_cvs. -- Tim Connors | Anglo-Australian Observatory http://site.aao.gov.au/twc | Coonabarabran, NSW 2357, Australia | Tel: +61 2 6842 6286
Subject: Re: [rt.cpan.org #30353] WITH_3D libGL tests
Date: Mon, 12 Nov 2007 04:11:15 +1100 (EST)
To: Chris Marshall via RT <bug-PDL [...] rt.cpan.org>
From: Tim Connors <twc [...] aaocbn.aao.gov.au>
On Mon, 12 Nov 2007, Tim Connors wrote: Show quoted text
> On Tue, 6 Nov 2007, Chris Marshall via RT wrote: >
> > Would you be willing to work from the CVS version of PDL > > instead of the current CPAN one? It would simplify getting > > fixes in for you and their migration into a 2.4.4 release by > > the end of this year.
> > Can do. > > http://pdl.sourceforge.net/FAQ > has login instructions to cvs.pdl.sourceforge.net, but it looks like > sourceforge has closed pserver access, so that could be out of date. BTW, > that faq has a bad link in the text > 3.3 ``CVS availability of PDL'' - should point to > http://pdl.sourceforge.net/FAQ/#q__4_8_can_i_get_pdl_via_cvs.
Woops, didn't realise I was replying to you personally. Have forwarded this onto jarle@astro.ox.ac.uk -- Tim Connors | Anglo-Australian Observatory http://site.aao.gov.au/twc | Coonabarabran, NSW 2357, Australia | Tel: +61 2 6842 6286
On Sun Nov 11 12:11:55 2007, twc@aaocbn.aao.gov.au wrote: Show quoted text
> On Mon, 12 Nov 2007, Tim Connors wrote: >
> > On Tue, 6 Nov 2007, Chris Marshall via RT wrote: > >
> > > Would you be willing to work from the CVS version of PDL > > > instead of the current CPAN one? It would simplify getting > > > fixes in for you and their migration into a 2.4.4 release by > > > the end of this year.
> > > > Can do. > > > > http://pdl.sourceforge.net/FAQ > > has login instructions to cvs.pdl.sourceforge.net, but it looks like > > sourceforge has closed pserver access, so that could be out of
date. BTW, Show quoted text
> > that faq has a bad link in the text > > 3.3 ``CVS availability of PDL'' - should point to > > http://pdl.sourceforge.net/FAQ/#q__4_8_can_i_get_pdl_via_cvs.
Hi, I did not get the replies and it has been awhile since I could check the CPAN RT site. Did you try cut and pasting the two lines for anonymous CVS access: cvs -d:pserver:anonymous@pdl.cvs.sourceforge.net:/cvsroot/pdl login cvs -z3 -d:pserver:anonymous@pdl.cvs.sourceforge.net:/cvsroot/pdl co -P PDL They worked for me orignally. I cannot verify from this account but your message has "cvs.pdl.sourceforge.net" while the above has "pdl.cvs.sourceforge.net". Maybe that is the problem. --Chris
Subject: Re: [rt.cpan.org #30353] WITH_3D libGL tests
Date: Thu, 29 Nov 2007 08:39:19 +1100 (EST)
To: Chris Marshall via RT <bug-PDL [...] rt.cpan.org>
From: Tim Connors <twc [...] aaocbn.aao.gov.au>
On Wed, 28 Nov 2007, Chris Marshall via RT wrote: Show quoted text
> > <URL: http://rt.cpan.org/Ticket/Display.html?id=30353 > > > On Sun Nov 11 12:11:55 2007, twc@aaocbn.aao.gov.au wrote:
> > On Mon, 12 Nov 2007, Tim Connors wrote: > >
> > > On Tue, 6 Nov 2007, Chris Marshall via RT wrote: > > >
> > > > Would you be willing to work from the CVS version of PDL > > > > instead of the current CPAN one? It would simplify getting > > > > fixes in for you and their migration into a 2.4.4 release by > > > > the end of this year.
> > > > > > Can do. > > > > > > http://pdl.sourceforge.net/FAQ > > > has login instructions to cvs.pdl.sourceforge.net, but it looks like > > > sourceforge has closed pserver access, so that could be out of
> date. BTW,
> > > that faq has a bad link in the text > > > 3.3 ``CVS availability of PDL'' - should point to > > > http://pdl.sourceforge.net/FAQ/#q__4_8_can_i_get_pdl_via_cvs.
> > Hi, I did not get the replies and it has been > awhile since I could check the CPAN RT site. > Did you try cut and pasting the two lines for > anonymous CVS access: > > cvs -d:pserver:anonymous@pdl.cvs.sourceforge.net:/cvsroot/pdl login > > cvs -z3 -d:pserver:anonymous@pdl.cvs.sourceforge.net:/cvsroot/pdl co -P > PDL > > They worked for me orignally. I cannot verify from this > account but your message has "cvs.pdl.sourceforge.net" > while the above has "pdl.cvs.sourceforge.net". Maybe > that is the problem.
It did indeed look like my problem. There was a further compilation problem that I can't remember the details of, but I'm on leave til next week so can't report back til later. -- Tim Connors | Anglo-Australian Observatory http://site.aao.gov.au/twc | Coonabarabran, NSW 2357, Australia | Tel: +61 2 6842 6286
See http://sourceforge.net/tracker/?func=browse&group_id=612&atid=100612 bug # 1994610 for follow up and resolution. Note the ticket subject was changed to reflect what is believed to be the issue here.
I cannot reproduce the error on my system. PDL::Graphics::TriD is being refactored so the OpenGL library generation will be changing. Further investigation of this issue will be tracked on the PDL sourceforge.net bug tracker accessible from http://sourceforge.net/projects/pdl/ Please report additional problems for this issue to the PDL bug tracker or via posting to the perldl mailing list. Thanks.
RT-Send-CC: tconnors [...] astro.swin.edu.au
Hi Tim- This is a final attempt to followup on your bug report from PDL-2.4.3 which is now more than two years old. We're up to PDL-2.4.6 and the TriD graphics have been refactored significantly. We use the Perl OpenGL module for the OpenGL bindings. I plan to close the PDL sf.net ticket for this issue unless I hear from you otherwise. If you get the chance to try the new PDL build using OpenGL, please let me know if there are any issues. Thanks in advance, Chris Marshall PDL/OpenGL developer
No follow up from the original reporter and the TriD build and implementation has been refactored to use the Perl OpenGL module and GLUT. The bug report is now out-of-date with respect to PDL-2.4.6 and forward. Please follow up with any further issues to the PDL bug tracker at sourceforge.net: http://sourceforge.net/tracker/?group_id=612&atid=100612 Thank you. --Chris
With the newly refactored PDL::Graphics::TriD code,
this issue is no longer relevant.  Setting it to resolved.

TODO: Check for any other pthreads issues in PDL.