mesa3d-dev

[Mesa3d-dev] [Bug 27254] New: glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254

           Summary: glBlitFramebuffer is defined while glBlitFramebufferEXT
                    is hidden as described by the XMLs
           Product: Mesa
           Version: unspecified
          Platform: Other
        OS/Version: All
            Status: NEW
          Severity: normal
          Priority: medium
         Component: Mesa core
        AssignedTo: mes...@li...
        ReportedBy: al...@au...
                CC: al...@au...


Hi all,
EXT_framebuffer_object.xml specifies that BlitFramebufferEXT to be hidden while
BlitFramebuffer is exposed as described in ARB_framebuffer_object.xml, is that
intentional? 

As GL_EXT_framebuffer_blit is supported since 6.5, I would have thought that
BlitFramebufferEXT should be defined in the library. 

It is unfortunately that when I test some of the functions using MESA, I need
to get a pointer to glBlitFramebuffer and when I test it on an older piece of
hardware then I need to get a pointer to glBlitFramebufferEXT.

Anyway, thanks so much for reading, I just want to know rather this is a bug or
the actual implementation so that I can do the appropriate thing in my program.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Mesa3d-dev] [Bug 27254] glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254





--- Comment #1 from Tom Fogal <tf...@al...>  2010-03-22 17:41:58 PST ---
> EXT_framebuffer_object.xml specifies that BlitFramebufferEXT
> to be hidden while BlitFramebuffer is exposed as described in
> ARB_framebuffer_object.xml, is that intentional?

This is reasonable, at least.  The Linux OpenGL ABI only requires
that OpenGL 1.2, GLU 1.3, GLX 1.3, and ARB_multitexture symbols be
available.

I refer you to the ABI, particularly section 3:

  http://www.opengl.org/registry/ABI/#3

See especially 3.4, 3.5, and 3.6.

> As GL_EXT_framebuffer_blit is supported since 6.5, I would have
> thought that BlitFramebufferEXT should be defined in the library.
>
> It is unfortunately that when I test some of the functions using
> MESA, I need to get a pointer to glBlitFramebuffer and when I test
> it on an older piece of hardware then I need to get a pointer to
> glBlitFramebufferEXT.

Sorry; this is how OpenGL is specified.  As you can read above,
exposing glBlitFramebuffer is an implementation courtesy.  An argument
could be made that nothing above GL 1.2 should be exported, so that
developers like you and I would more quickly ascertain that we are
writing invalid code!  That'd probably be a bad idea though, because
a lot of `in the wild' code is likely to break if we remove all those
entry points.

Anyway, the only valid way to get glBlitFramebuffer, or any extension
really, is to load it dynamically.  This is why I recommend GLEW :)


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Mesa3d-dev] [Bug 27254] glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254





--- Comment #2 from Alan <al...@au...>  2010-03-22 18:49:55 PST ---
(In reply to comment #1)
> > EXT_framebuffer_object.xml specifies that BlitFramebufferEXT
> > to be hidden while BlitFramebuffer is exposed as described in
> > ARB_framebuffer_object.xml, is that intentional?
> 
> This is reasonable, at least.  The Linux OpenGL ABI only requires
> that OpenGL 1.2, GLU 1.3, GLX 1.3, and ARB_multitexture symbols be
> available.
> 
> I refer you to the ABI, particularly section 3:
> 
>   http://www.opengl.org/registry/ABI/#3
> 
> See especially 3.4, 3.5, and 3.6.
> 
> > As GL_EXT_framebuffer_blit is supported since 6.5, I would have
> > thought that BlitFramebufferEXT should be defined in the library.
> >
> > It is unfortunately that when I test some of the functions using
> > MESA, I need to get a pointer to glBlitFramebuffer and when I test
> > it on an older piece of hardware then I need to get a pointer to
> > glBlitFramebufferEXT.
> 
> Sorry; this is how OpenGL is specified.  As you can read above,
> exposing glBlitFramebuffer is an implementation courtesy.  An argument
> could be made that nothing above GL 1.2 should be exported, so that
> developers like you and I would more quickly ascertain that we are
> writing invalid code!  That'd probably be a bad idea though, because
> a lot of `in the wild' code is likely to break if we remove all those
> entry points.
> 
> Anyway, the only valid way to get glBlitFramebuffer, or any extension
> really, is to load it dynamically.  This is why I recommend GLEW :)
> 

I agree with what you said and have planned to use GLEW, this gives me extra
motivation to use it ^_^ 
I guess I am just curious why this one is left out as all the other GL calls we
use in our software are included in Mesa's libGL, especially when
BlitFramebuffer is just an alias to it.

Anyway, thanks a lot. Once we get GLEW going here then I probably won't care
much about it.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Mesa3d-dev] [Bug 27254] glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254





--- Comment #3 from Ian Romanick <id...@fr...>  2010-03-23 00:09:37 PST ---
(In reply to comment #2)
> I guess I am just curious why this one is left out as all the other GL calls we
> use in our software are included in Mesa's libGL, especially when
> BlitFramebuffer is just an alias to it.

I think that's a bug.  Our plan of record is not provide static symbols for
*any* new functions.  If there is a static symbol available for
glBlitFramebuffer (or any other function added in the last 4 years or so), it
should be removed.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Mesa3d-dev] [Bug 27254] glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254





--- Comment #4 from Chia-I Wu <ol...@gm...>  2010-03-23 00:38:57 PST ---
I think I should brought this up since 7.8 is about to be released.  In
gl_API.xml, we have

  <function name="HistogramEXT" alias="Histogram" static_dispatch="false">

C and X86-64 dispatcher correctly hide glHistogramEXT.  However, up until
recently, X86 dispatcher exposed the function.  I've made a commit

  41a87a43e11c664935349f938022d58d3e22da4e

to honor gl_API.xml and hide the symbol (and some others).  Does this sound
fine to your or is this considered an ABI breakage?


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.

[Mesa3d-dev] [Bug 27254] glBlitFramebuffer is defined while glBlitFramebufferEXT is hidden as described by the XMLs

[<bug...@fr...>]

http://bugs.freedesktop.org/show_bug.cgi?id=27254


Ian Romanick <id...@fr...> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |NOTABUG




--- Comment #5 from Ian Romanick <id...@fr...>  2010-03-26 09:49:15 PST ---
(In reply to comment #4)
> I think I should brought this up since 7.8 is about to be released.  In
> gl_API.xml, we have
> 
>   <function name="HistogramEXT" alias="Histogram" static_dispatch="false">
> 
> C and X86-64 dispatcher correctly hide glHistogramEXT.  However, up until
> recently, X86 dispatcher exposed the function.  I've made a commit
> 
>   41a87a43e11c664935349f938022d58d3e22da4e
> 
> to honor gl_API.xml and hide the symbol (and some others).  Does this sound
> fine to your or is this considered an ABI breakage?

Sorry for the slowness of my reply.  As long as the set of symbols exposed on
x86 isn't reduced to less than the set of symbols exposed on x86-64, I think
the change should be fine.

Since the original report wasn't actually a bug in Mesa, I'm going to close
this as NOTABUG.


-- 
Configure bugmail: http://bugs.freedesktop.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.