On 28.09.2010 13:21, Jeff Trawick wrote:
On Tue, Sep 28, 2010 at 6:56 AM, Rainer Jung<[email protected]> wrote:
On 28.09.2010 12:44, [email protected] wrote:
Author: rjung
Date: Tue Sep 28 10:44:32 2010
New Revision: 1002110
URL: http://svn.apache.org/viewvc?rev=1002110&view=rev
Log:
Fix test suite for 2.0.x:
- Skipping test for PR 46428 when version< 2.2.12
(whitespace in proxied URL).
Does that type of change make sense?
I was running the test suite for 2.0 and it is failing at a couple of
places. As far as I can see until now, most (all?) of them are because the
corresponding patches haven't been backported to 2.0.
Does it make sense to deactivate the failing tests for all versions lower
than the backport version, or do people think it is better to continuously
nag about missing backports? Or maybe add a friendly reminder in the skip
message?
I think it is better to see messages about bugs present in the level
being tested.
Some reminder somewhere would be useful. I don't know what makes the
most sense. Even an entry in the httpd-test/STATUS or perhaps
2.0.x/STATUS file about test failures that have been diagnosed (and
can be skipped if you're just looking for regressions) would be
useful.
Known test failures:
2.0.64-dev as of rXXXXXXX:
PR46xxx test fails because fix hasn't been backported to 2.0.x branch
ACK.
I finally found the "todo" feature ánd used it depending on the tested
Apache version. I also added explanatory STDERR message to the failing
tests.
Since all of those only needed simple backports, I added them as
proposals to the 2.0 STATUS file.
Rainer
