This is the mail archive of the
libstdc++@gcc.gnu.org
mailing list for the libstdc++ project.
Re: *_data_map_rand.cc timeout on mips
- From: Mark Mitchell <mark at codesourcery dot com>
- To: Benjamin Kosnik <bkoz at redhat dot com>
- Cc: Ami Tavory <atavory at gmail dot com>, libstdc++ at gcc dot gnu dot org
- Date: Fri, 07 Jul 2006 18:01:50 -0700
- Subject: Re: *_data_map_rand.cc timeout on mips
- References: <20060620213922.GA14910@deprecation.cyrius.com> <e00e7abf0606220824mc3d32d6j34af87f4854990b8@mail.gmail.com> <e00e7abf0606220835t33f4068cw97c46935a6fefc52@mail.gmail.com> <20060629125558.d141fe99.bkoz@redhat.com>
Benjamin Kosnik wrote:
> In addition, this would mean that testresults on the same machine would
> not be the same, run after run. Since we're going for consistency, this
> would be a problem.
I agree. I do not think randomization is a good idea in our testsuites,
even though I do understand the fact that randomization helps to cover
the test space over time. If we were to do that, the name of the test
should include the random variable seed so that, at least, we do not
have the same test passing one time and failing the next.
--
Mark Mitchell
CodeSourcery
mark@codesourcery.com
(650) 331-3385 x713