Loop unroll fixes

[Jim Wilson]

>	Zoltan repeatedly posted messages at various intervals asking
>about the patch.

I addressed this in my first message on this thread, with documentation.
If you wish to contest this, then please make specific objections.
My message is at:
http://gcc.gnu.org/ml/gcc/2001-09/msg00533.html

To recap, Zoltan posted the patch in July, 6 people got involved in the review,
3 people made some claim to having fixed it, and then it died, apparently
because everyone thought someone else was handling the problem.  At the time,
I had sufficient reason to believe that someone else had already fixed the
problem.

Zoltan brought up the issue again in August.  The issue was deliberately
postponed because it conflicted with the gcc 3.0.1 release.

Zoltan brought up the issue again in September.  At this point, you claimed
drastic action was needed.

I see no public evidence that the patch was ignored by anyone, and hence no
support for your attempt to force the issue.  The two month delay is
unfortunate, but understandable given the circumstances.

>My private communication only followed up after each of
>his repeated public requests.

I claim that this was a mistake.  Your use of private communication hid the
fact that there was a problem from the other gcc developers.  If the
followups had been posted to one of the public gcc lists, then others
would have known about the problem, and likely someone would have volunteered
to help.

For example, in one of the first messages I got in July, you indicated that
you and Mark Mitchell would check in the patch.  It was only yesterday, that
I discovered that you did not check in the patch because Bernd said he wanted
to review it.  If that fact had been mentioned on a public mailing list, or
in the PR, then I would have known that the issue had not been resolved yet.
But it appears that the only place this fact was mentioned was in private
communications with others.  Hence it is not surprising that I did not realize
that the issue was still unresolved.

>	I do not accept your statement that the patch and its lack of
>attention was secret.

I never stated that the patch was secret.  I did state that the lack of
attention was secret.  You have produced no evidence to indicate otherwise.
If this claim is wrong, then please point me substantiating evidence on
the gcc.gnu.org site.  I believe I have produced suffficient evidence to
indicate otherwise.  See my mail message referenced above.

>  I feel like you are trying to blame the victim.
>Maybe that's just your reaction to my forcing the issue after months of
>frustration. 

I am not trying to blame you for the problem.  I am trying to help you.
But I can not help you unless you admit that you made a mistake by relying
too much on private email, and by using inappropriate tactics to try to
force the issue.

Yes, you were frustrated, but no one knew that you were frustrated, and
thus no one was able to help you.  To fix that problem, you need to send
messages to public mailing lists instead of sending only private mail, or
else put an audit-trail in the PR.  That way, everyone is aware of what is
going on.

Also, you conveniently left out the fact that your original mesasge made
veiled accusations against unnamed targets with no supporting evidence.
I don't like to see people's reputations unfairly tarnished, particularly
when the accusations were ambiguous enough that people might think that I
personally was at fault.  This action was definitely not supportable by
the public evidence.

But to get back to the real problem here, the lack of patch review for
Zoltan's patch... Bernd is already reviewing it as we speak, and I have
already stated that I am willing to review the patch, so I will follow up
if Bernd doesn't finish resolving the issue.

As for the larger issue of patch review in general, Joe Buck has started
a new thread to discuss it.  A thread that is largely free of invective.
I hope it stays that way, and I hope that some good ideas come out of it.

Jim