This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: automatically updating ulps -- any review needed ?
- From: "Joseph S. Myers" <joseph at codesourcery dot com>
- To: Andreas Schwab <schwab at linux-m68k dot org>
- Cc: Mike Frysinger <vapier at gentoo dot org>, <libc-alpha at sourceware dot org>
- Date: Mon, 6 Jan 2014 15:38:46 +0000
- Subject: Re: automatically updating ulps -- any review needed ?
- Authentication-results: sourceware.org; auth=none
- References: <201401051711 dot 28152 dot vapier at gentoo dot org> <Pine dot LNX dot 4 dot 64 dot 1401061409090 dot 16360 at digraph dot polyomino dot org dot uk> <87r48lgo4b dot fsf at igel dot home>
On Mon, 6 Jan 2014, Andreas Schwab wrote:
> "Joseph S. Myers" <joseph@codesourcery.com> writes:
>
> > Note that the checked-in file should be truncated before running "make
> > regen-ulps", so as to remove old ulps for tests for which errors have gone
> > down or whose names have changed.
>
> Shouldn't regen-ulps run the tests with --ignore-max-ulp=yes?
That would only work if the ulps were independent of the compiler and
hardware in use. For x86 and x86_64, for example, regeneration for a
release involves one from-scratch regeneration after truncation, then
subsequent incremental use of regen-ulps by people who see ulps because of
different compilers or hardware in use (and similarly, outside of the
freeze, from-scratch x86 / x86_64 regenerations would unnecessarily
introduce ulps failures for people with hardware different from that used
for the last regeneration).
--
Joseph S. Myers
joseph@codesourcery.com