Alex Romosan wrote:

> "Frederic Bouvier" <[EMAIL PROTECTED]> writes:
> 
> > Anyway, a fix is in CVS now
> 
> not exactly. the "fix" is ugly beyond belief. it basically boils down
> to commenting out dlclose() (despite all that code being moved
> around). it also misses out a call to dlerror() after dlopen() and
> before dlsym() to clear the error. anyway, we are not getting anywhere
> here as erik has some strange beliefs about how shared libraries work.
> his comment on checking in the hack: "Work around a broken
> dlopen/dlclose Linux implementation", speaks volumes...

I was not understanding Erik point of view of advocating a patently 
broken code but now I am really missing your point :-(

This code now dlopen libGL.so only once and ensures that the *same*
handle is always used instead of using a different one every time.
We know that the library has to be linked until the end of the 
program so it would be over-engineering to bother closing that 
handle the OS will close anyway.

Also I don't see the point to clear an error on dlopen. If the library
is not here, the program cannot run anyway.

-Fred



_______________________________________________
Flightgear-devel mailing list
[EMAIL PROTECTED]
http://mail.flightgear.org/mailman/listinfo/flightgear-devel

Reply via email to