Re: [Plib-users] Library Initialization

[Steve B. <sjb...@ai...>]

Elad Lahav wrote:
> 
> OK, I have tried to exclude as many reasons as possible for this failure,
> and this is what I got:
> 1. Reinstalled RH 7,

*BZZT*BZZT* WARNING! WARNING!

You might want to downgrade your gcc to something that works.  RedHat
are coming in for a lot of flak over their decision to ship a relatively
poorly tested beta version.

> with the original kernel and libraries, and still got
> the same GL context issue...

Well - I'm rather unconvinced - I don't think this is a PLIB problem.

> 2. Compiled other, non-plib, GL-dependant programmes. They seemed to work
> fine.

Well, that's interesting - but I *still* don't think it's a PLIB problem.

The deal is that you really havn't run much PLIB code by this stage:

The rule is that the first SSG function to be called MUST be ssgInit - and
ssgInit must not be called until there is a valid OpenGL rendering context.

The very first line of that function is a call to glIsValidContext whose
very first line is to ask X if there is a valid rendering context.
Immediately upon return, you are getting that error message saying
that there isn't a valid context.

So, if FGFS has a valid rendering context in the line immediately before
the very first ever call to a PLIB function (well, a PUI, FNT or SSG function
at least) - then you could insert a glXGetCurrentContext in the line just
before ssgInit is called and it would return some non-NULL value.

Then you are telling me that sometime in the next three lines, PLIB somehow
corrupts that context and then - testing it a second time - get's NULL back?

I don't believe that.

Ergo FGFS did not have a valid context before calling ssgInit - so it's
FGFS's fault.  I can't see a flaw in that reasoning.

You may not previously have seen this message with earlier versions
of PLIB because we were not doing the check - now we are, we may have
uncovered a FGFS error of some kind....I don't know.

> 3. When trying FlightGear, I have compiled all libraries and programmes
> (plib, SimGear and FlightGear), on the same machine, with the same compiler,
> at the same time (so you can't even blame it on weather changes...;-)
> Anything else?

Well, I think you should hack into plib/src/ssg/ssg.cxx, see where it says 'glXGetCurrentContext'
in 'glIsValidContext' - print out the value that glXGetCurrentContext returns.  Also stick
a printf inside 'ssgInit' so we know that it is truly what called glIsValidContext.

Now, put a check for glXGetCurrentContext into FGFS in the line before you call
ssgInit - and you'll know for sure.

Incidentally, a similar test for a valid OpenGL context is made in PUI's puInit
(which FGFS is presumably also calling)...and also in the FNT font loader. 
*PRESUMABLY* those are being called after ssgInit - so we don't see them complain.

I *suspect* the RedHat C++ compiler - it's been implicated in several problems in
other projects this week.  If I'd bought a copy of RH 7, I'd be demanding my money
back.  They did a VERY irresponsible thing and use poor package maintainers are
suffering the consequences.  That gets me somewhat upset.  I can only imagine
what the GCC maintainers are going through.


-- 
Steve Baker   HomeEmail: <sjb...@ai...>
              WorkEmail: <sj...@li...>
              HomePage : http://web2.airmail.net/sjbaker1
              Projects : http://plib.sourceforge.net
                         http://tuxaqfh.sourceforge.net
                         http://tuxkart.sourceforge.net
                         http://prettypoly.sourceforge.net