This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [RFC][PATCH 22/22] Add Infinity notes implementing td_ta_map_lwp2thr
- From: Gary Benson <gbenson at redhat dot com>
- To: Szabolcs Nagy <szabolcs dot nagy at arm dot com>
- Cc: Adhemerval Zanella <adhemerval dot zanella at linaro dot org>, libc-alpha at sourceware dot org
- Date: Thu, 22 Oct 2015 14:31:50 +0100
- Subject: Re: [RFC][PATCH 22/22] Add Infinity notes implementing td_ta_map_lwp2thr
- Authentication-results: sourceware.org; auth=none
- References: <1445352975-17844-23-git-send-email-gbenson at redhat dot com> <56277179 dot 5060502 at linaro dot org> <20151021144250 dot GA18821 at blade dot nx> <5627BEBA dot 5040100 at linaro dot org> <20151021190911 dot GB25328 at blade dot nx> <5627ED06 dot 5040505 at linaro dot org> <20151021212857 dot GA10785 at blade dot nx> <F688AA9A-97E8-4271-817A-C4F4E0201D78 at linaro dot org> <20151022085645 dot GA21239 at blade dot nx> <5628BFBB dot 6060706 at arm dot com>
Szabolcs Nagy wrote:
> On 22/10/15 09:56, Gary Benson wrote:
> > Don't people usually cross-build to bootstrap new platforms?
>
> the easiest way to run the glibc tests is to build it natively
> (because the test system has assumptions about the host env that
> is hard to match with the ssh+nfs cross test setup).
Ok.
> currently the test system depends on perl (for header conformance
> tests and memory leak tests) which is already a pain.
Ok.
> python is another hard to build dependency...
Granted. But... is this something you have actually tried, or are
you basing this on your experience with Perl? Some years ago I
worked on packaging a bunch of things (including Python and Perl)
for various free and proprietary systems and Perl was always the
time-sink. Funnily enough it was there to support a test suite
then too...
> ...and you have to make sure that python language changes don't
> affect the behaviour.
For sure. Both I8C and I8X have pretty comprehensive test suites.
I8C has tests with 98% coverage so I'm pretty sure it's doing what
it should. I test every commit on Pythons 2.[67], 3.[1-5] and the
latest development and nightly snapshots. I8X has only 87% coverage
but I'm working on that. I'd like both to be 100% by the time I
have a full GLIBC series ready to commit.
> i think this is not a critical issue since the build can succeed
> without i8*, but it is worth noting.
Yes, if the tools aren't installed the build will proceed as normal,
you just won't get the extra stuff this series adds.
Thanks,
Gary
--
http://gbenson.net/