This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: Linux Foundation Collaboration Summit: The future of The GNU CLibrary.
On Wed, 28 Mar 2012, Mike Frysinger wrote:
> > If someone wants to work on the testsuite infrastructure I'd advise
> > several smaller steps before actually setting up any general test harness.
> > In particular:
>
> whatever happened to qmtest ?
https://github.com/MentorEmbedded/qmtest
Whatever the approach, it's still important to split up changes as far as
possible into minimal self-contained reviewable units, and avoid tying
bugfixes unnecessarily to changes to use a test harness. So all the
separate steps I suggest would be relevant regardless. They are also
comparatively noncontroversial - whereas it's evident from this discussion
that the choice of test harness software is more controversial and would
need more careful and extended consideration.
Ideally, a change to make the code use a test harness would not change at
all the commands and environments in which any test is run, or the set of
tests run; it would just mean that the tests start running under the
harness instead of directly from makefiles, but otherwise run the same way
as before. This means it would be a good idea to work out a way to make
the metadata about tests usable in both ways simultaneously, so that
introducing a harness does not require large and difficult-to-review
changes moving all the metadata from one place to another. Maybe if tests
can be described purely by makefile variables, without custom rules, it
would be possible for makefile code to generate files with test details in
a suitable form for use with a separate harness.
(That's just one possibility. As noted recently regarding the build
system <http://sourceware.org/ml/libc-alpha/2012-03/msg00958.html>, major
changes like this require careful thought and discussion and incremental
improvement, not code dumps of months of work; my suggestion is of one way
it might be possible to get quite close to using a test harness by way of
a series of incremental patches that can be seen to be individually
desirable independent of what test harness may end up being used.)
--
Joseph S. Myers
joseph@codesourcery.com