Re: [Call for middle-end reviewer] Oustanding visibility PR 20218

[Roger Sayle <roger at eyesopen dot com>]

On Mon, 23 May 2005, Paolo Carlini wrote:
> Only a quick comment, however, not of technical nature, a clarification:
> actually my remark above was partially *wrong*, in this sense: I
> confused submitter of 20218 (HJ himself) with submitter of 19664 (Pawel
> Sikora, "pluto"): the latter indeed reported to have tested the entire
> set (19664 + 20218) on other arches beyond those three: I see now that
> probably he didn't run the entire testsuite. Sorry about the confusion.

Ahh!  Thanks for the clarification.  I thought it a bit odd when you
referred to the bug submitter, when for 20218 that was the same person
as the patch author.

Looking through PR19664, it looks like Andreas has indeed additionally
tested on powerpc, alpha and sparc.  By great luck/misfortune, these
additional platforms are all unaffected by the bug in HJ's patch, so
even full bootstraps and regression tests won't expose the regressions.

My apologies also for the delay in replying.  Neither 20218 nor
19664 are technically regressions, the workaround is to avoid the
new version of binutils which exposes this problem.  So I'd assumed
that I could wait until the unslush, and by then have test on
IA64/HPUX to ensure nothing breaks before approving.

Unfortunately, re-reading through the changes whilst waiting for the
IA-64 bootstrap and regression test to finish exposed deeper issues.

Thanks for your clarification, sorry for my confusion.

Roger
--