[Cuis] 4.1 to 4.2 migration question

[Phil (list)]

On Mon, 2015-02-16 at 20:37 -0800, Casey Ransberger wrote:
> If memory serves (and it may not) Igor mentioned having automatically generated bindings based on the API at one point when I asked about it. The comment in there made me suspect that you might be adapting generated code that he'd had his hands in. Might be worth asking him. 
> 

I think the code I'm currently using was indirectly the result of the
Croquet work.  I believe it was a package that Andreas Raab originally
generated (the code to do so was lost to the sands of time way back when
I had asked about it)  The package had been last touched by John Dougan
and I merged in some later fixes from Bert Freudenberg.

> One reason I'm interested here is, the Croquet stuff isn't just tangled up with a synched messaging model, it's also written in a style of GL that doesn't really seem to be supported anymore. So automatically generating bindings for something like e.g. OpenGL ES on the Pi is really compelling right now. 
> 

Tell me about it... I had to untangle that crap before I eventually
found a Squeak package for OpenGL that worked.  Yes, this is old
immediate mode code.  I had planned to switch over to a subset of OpenGL
(i.e. OpenGL ES as supported by OpenGL 4.1 iirc) but haven't had the
right combination of time and need.  If I do (before someone else does
it), I'll definitely be looking to regenerate it from the OpenGL XML
spec files from scratch rather than modifying what I currently have.

Unfortunately, most of the more recent OpenGL work by others appears to
be happening in Pharo and was based on Alien, now NativeBoost (next
month probably something else)... so on top of porting the OpenGL code,
you've got to port the code that it depends on (which in turn often
depends on yet more stuff) plus dealing with another plugin.  Feel free
to do this if you want :-)  I was just starting to sort out Alien when
NB came along and... I got weary of chasing these undocumented moving
targets, hence the older code I'm using.

> One comment inline. 
> 
> > On Feb 16, 2015, at 8:09 PM, "Phil (list)" <pbpublist at gmail.com> wrote:
> > 
> >> On Mon, 2015-02-16 at 19:26 -0800, Casey Ransberger wrote:
> >> What is self parserClass expected to answer when one happens to be an
> >> instance of OGLExtManager?
> > In Cuis 4.1 it answers with Parser
> 
> Hmm. And that worked with this stuff in 4.1? Seems odd to explicitly call out the default parser. Might be worth looking at what comes back in Squeak and (gut says) Pharo to be sure. 
> 

Yes.  I was originally using it in Squeak 3.x / Pharo 1.x (I think... it
might have been Pharo 2.x) and it's been working in Cuis since 1.x.

For the time being I've just gone and used Parser directly which works
around the immediate error so I can see what else needs fixing.