[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20130410134051.GC20323@brightrain.aerifal.cx>
Date: Wed, 10 Apr 2013 09:40:51 -0400
From: Rich Felker <dalias@...ifal.cx>
To: musl@...ts.openwall.com
Subject: Re: Test suite for musl
On Wed, Apr 10, 2013 at 01:14:07PM +0200, Daniel Cegiełka wrote:
> 2013/4/10 Timerlan Moldobaev <moldobaev@...il.com>:
> > Hi ,
> >
> > Is there any test suite besides libc-bench that was used to check various
> > aspects of musl like functional correctness and the compliance to POSIX
> > format ?
>
> yup:
>
> http://git.musl-libc.org/cgit/libc-bench/
> http://git.musl-libc.org/cgit/libc-testsuite/
There are also:
Open POSIX Test Suite
http://posixtest.sourceforge.net/
This one is a bit outdated (not updated to POSIX 2008) and contains a
number of incorrect or invalid tests, mostly in the form of tests that
invoke undefined behavior. Unfortunately we have not compiled I list
of those tests to remove/disable them, but somebody could probably do
that... [This would be a great task for anybody who wants to help!]
CLUTS
https://github.com/lmarcetic/cluts
This one was developed early on with testing musl in mind as a GSoC
project under Openwall, but unfortunately it has not been maintained,
probably since the test framework was never completed. Still it has
some useful tests.
nsz's test repo
http://nsz.repo.hu/git/?p=libc-test
A lot of the code here is based on libc-testsuite, but there is also a
good deal of additional floating point test code and maybe a few other
things.
gnulib tests
These are widely available in packages that use gnulib, or from
upstream. Some are used in configure scripts (these will quickly catch
regressions that cause packages to start using bloated gnulib
replacements for functions) and others are heavier weight and only
invoked as part of test targets.
Finally, I have a few more individual test modules (mainly for
strtod/scanf/etc.) which are not in any test package. I believe I've
sent these to the list a few times before, but I could make them
available at some point separately. Unfortunately they are not
integrated into any test framework.
Developing a comprehensive test framework for conformance and
regressions has been a goal for a long time, but unfortunately nobody
has stepped forward to do it and it's not something I or the other
core developers have time to do.
Rich
Powered by blists - more mailing lists
Confused about mailing lists and their use?
Read about mailing lists on Wikipedia
and check out these
guidelines on proper formatting of your messages.
