apr-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Karl Fogel <kfo...@newton.ch.collab.net>
Subject Re: cvs commit: apr/test testfile.c
Date Thu, 02 Jan 2003 15:39:45 GMT
<rbb@apache.org> writes:
> There are two ways to resolve this bug.  You can create a feature macro
> APR_FILE_OPEN_DIRS, which means that apr_file_open will work on
> directories, or you can make apr_file_open fail to open directories on
> Unix.  In the meantime, leave the failure in the test suite.  It is
> highlighting a problem.

I'm not very familiar with the APR test framework, so pardon me if the
answer to this is obvious: do we have an XFAIL mode available?  If so,
we should be using it here.

If a test failure reveals a bug, the bug should be recorded in the
appropriate place (um, "the issue tracker"... :-) ).  If we then
switch the test to "eXpected Fail", that reduces the noise level so
that test won't interfere with our seeing other failures.  (And in the
issue, just record that there is a test, and that it needs to be
changed from XFAIL to PASS when the bug is fixed.)

I agree that removing a failing test of a valid feature is bad; but
using the test suite as bug tracking system is also bad.

Constant test failures won't "remind" us to fix bugs -- they'll just
acclimatize us to seeing test failures, until they're considered
normal.  We have to keep the failures special, by making them
exceptional events worthy of immediate action, even if that action is
just to record the bug in some canonical place.

-Karl

Mime
View raw message