Re: [Plib-devel] Making a PLIB release.

[Bert D. <dri...@pl...>]

On Tue, 11 Jan 2005, Steve Baker wrote:

> I think it's time we made another PLIB release.
>
> There have been quite a few bug fixes and minor enhancements
> since last time.
>
> Unless anyone has any burning issues, I'll cut a release from
> the current CVS tomorrow evening (that's Wednesday 11th Jan).

I just applied the two patches from the FreeBSD ports tree that are 
required to unbreak the build on FreeBSD 5.3. The patch to jsBSD.cxx is 
ugly and I'd have rewritten it if I had the couple of hours required to 
do it in autoconf and verify the change at least won't break Linux. But 
the patch works and it does unbreak the build, and it's isolated so it 
won't break other platforms. I've made a mental note to clean it up when 
I'm bored.

I did not apply the patch to src/ssg/ssgDList.cxx; it adds a large chunk 
of code to implement an _ssgMatrixStack class and I've got no clue what 
it's for -- any takers?

As Martin observed, the patch for netSocket.cxx was already in plib for 
quite a while.

Counter to popular belief, FreeBSD 5.3 does not fix the pthread issue. 
I've narrowed it down a bit. I've installed a plain-jane build of cvs 
plib with --prefix=/usr/local/cvs-plib . When I build ttt3d with

   #!/bin/sh
   prefix=/usr/local/cvs-plib
   env LDFLAGS="-L$prefix/lib" \
       CPPFLAGS="-I$prefix/include" \
 	./configure --prefix=$prefix --with-GL=/usr/X11R6
   gmake

the resulting binary dies with
   FATAL: ssgInit called without a valid OpenGL context.

When I configure ttt3d with
   #!/bin/sh
   prefix=/usr/local/cvs-plib
   env LDFLAGS="-pthread -L$prefix/lib" \
       CPPFLAGS="-pthread -I$prefix/include" \
 	./configure --prefix=$prefix --with-GL=/usr/X11R6
   gmake

it runs just fine.

The good news is that (a) the FreeBSD port's patch to take out the check 
is not needed, (b) we probably have an answer to people who claim it is, 
and (c) that a default build of plib works fine if the app is linked 
properly (I can only surmise that it's the linking with -lc_r that is 
implicit in compiling with -pthread that "fixes" it; I've still got no 
clue what actually _causes_ the problem).

Anyway, when all is said and done, I think we're golden. I'm pretty 
convinced that I didn't break anything, but I'll be watching my mailbox 
like a hawk during the release period.

Cheers,

 				-- Bert

-- 
Bert Driehuis -- dri...@pl... -- +31-20-3116119
If the only tool you've got is an axe, every problem looks like fun!