[PATCH 2.95.3 - WITHDRAWN] Fix powerpc-wrs-vxworks eabi backwards incompatibility.

[David Korn]

>	The DFmode alignment issue was discussed among the GCC PowerPC
>target maintainers, PowerPC developers, and others.  We agreed at the end
>of the discussion (which extended over a year) to correct the
>implementation of the ABI in the compiler and that vendors would shoulder
>the responsibility of compatibility. 

>	We can revisit this decision, but I am not confident that the
>discussion would be useful.

  Nope, I can't see any point in that either. I didn't come across any
trace of the discussion about this in the ml archives, but there you go;
I wasn't looking for it, because I didn't know it was there.

>	Free Software is exactly what gives you the ability to create and
>disseminate patches to provide backward compatibility with gcc-2.7.2.  I
>encourage you to make them widely available.  I encourage you and WRS and
>Red Hat to work together on this compatibility functionality if it is
>useful to any of you.  The patches are very helpful to a group of people,
>but not every useful patch should be incorporated into the public GCC
>sources. 

  I agree entirely; I've already half-finished building the website on
which I was intending to release these patches regardless of the decision
reached here. I brought the subject up here for two reasons: to get advice
from people with more experience working on Gcc than me about the best
approaches (target flag? attribute? new target triple?), and to find out if
the Gcc team considered these patches worthwhile for the central repository.
I have my answer to the second of those, but I'd still be interested in
any opinions anyone has to offer about the first.

>	I would like to see the CALL_V4_SET_FP_ARGS patch separated out.
>If the current implementation has a bug in its use of those macro names,
>it should be fixed.  That part of the patch may fix a bug (at least a
>documentation bug) in the current development sources.

  Will do. I'll send patches against 2.95-branch and head of tree v3 either
tomorrow or the day after. They should qualify as obviously correct, given
that all the patch involves is exchanging ..CLEAR.. and ..SET.. everywhere
they occur (except in the #defines which give them numeric values).

  In the light of the already-decided gcc policy and the nearness of
the deadline, I would like to ask Bernd if he would accept the patches
for -mlongcall and -mvxcompat272 as donations for the contrib/ subdir?
I won't ask any more for them to be included in the main code now that the
deadline is so close, and I can't see any point in developing the vxcompat
patch for v3.0, although I'd be happy to work up the longcall patch for it.

      DaveK
-- 
The Boulder Pledge: "Under no circumstances will I ever purchase anything 
offered to me as the result of an unsolicited email message. Nor will I 
forward chain letters, petitions, mass mailings, or virus warnings to large 
numbers of others. This is my contribution to the survival of the online
community." 


**********************************************************************
This email and any files transmitted with it are confidential and
intended solely for the use of the individual or entity to whom they
are addressed. If you have received this email in error please notify
the system manager.

This footnote also confirms that this email message has been swept by
MIMEsweeper for the presence of computer viruses.

www.mimesweeper.com
**********************************************************************