Re: [Bug target/10129] [3.4 only] Ada bootstrap fails on PPC-Darwin- invalid assembler emitted - PIC related

[Mark Mitchell <mark at codesourcery dot com>]

Geert Bosch wrote:

> It's really unfortunate as this bug is a regression reported
> more than a year ago. September last year, the bug had been fixed,
> but the fix went unreviewed and not approved for 3.3 as "Ada is not
> release critical".
...
> Basically, the message seems to be that if Geoffrey Keating breaks
> Ada bootstrap that's fine. However, fixes to restore bootstrapping
> cannot go in for another two major releases? Please explain.

No, that's not the situation.  Here are my thoughts:

(1) Assuming that it was Geoff who broke the bootstrap (and I have not 
investigated, so I cannot say) then he has a responsibility to fix the 
problem.  The best mechanism we have is to make sure that Geoff knows 
about the problem and demonstrate that the problem was caused by his 
patch.  Usually, when it's obvious that a particular patch caused the 
problem, the person who committed that patch feels obligated to fix the 
problem, as that is the implicit oglibation of patch committers.

If you feel that a particular person is not doing enough to fix a 
problem that he or she caused, then you should contact that person, and, 
if absolutely necessary, contact the Steering Committee.

In this particular case, I don't see in the PR that anybody attributed 
the problem directly to one of Geoff's patches, so it's entirely 
possible that even if his patch was the cause, he had no knowledge of it.

(2) If there is a patch that solves the problem, then it would be good 
if that patch could be reviewed and committed.  However, there are some 
patches that I do not have the knowledge to review, and I have limited 
time; I do not consider it my obligation to review all patches.  If youi 
feel a patch is not being reviewed, the best thing you can do is contact 
the maintainers of that part of the compiler and ask them for a review, 
or, if absolutely necessary, contact the Steering Committee.

In Comment #30 in the PR trail, I specifically said the patch was OK for 
3.4.x if a Darwin maintainer would review it.  Apparently, that hasn't 
happened.  Did you contact them directly to ask for a review?

(3) The fact that I'm not holding up the release for this PR doesn't 
mean that I'm rejecting the patch; it just means that I'm not going to 
hold up the release for it.

--
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com