Jani Averbach wrote:
>Hello,
>
>I think it would be nice if svntest won't run tests when there isn't
>anything to test. This is the case for branches most of the time.
>
>But I don't know how other people like to use and read svntest.sh's
>emails, so I am asking here what you think about this change.
>
>Here is an example what the output would be when testing is no-op.
>
>http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=12005
>http://subversion.tigris.org/servlets/ReadMsg?list=svn-breakage&msgNo=12004
>
>BR, Jani
>
>Log:
>The svntest framework won't run tests if there aren't any changes to test.
>
>* tools/test-scripts/svntest/svntest.sh
> Abort testing if all rebuild flags are zero.
>
>* tools/test-scripts/svntest/svntest-update.sh
> (SVN_UPDATE_REGEXP_1, SVN_UPDATE_REGEXP_2): New variables
> (UPDATE_REBUILD_FLAG): New parameter 'mode', this func. could now
> check rebuild status for SVN projects too.
>
>* tools/test-scripts/svntest/svntest-rebuild.sh
> Added updating of SVN's rebuild flag.
>
>* tools/test-scripts/svntest/svntest-sendmail.sh
> Added new email target 'NOOP'.
>
>
The idea itself is nice, but the patch itself assumes that you have one
copy of apr-*, httpd and svn for testing. In my case, I use a single
copy of apr and httpd with three different svn working copies (trunk,
1.0.x, 1.1.x); so with your patch, if I run the tests in that order and,
e.g., APR gets updated during the trunk tests, but not during the 1.0.x
tests, and there are no changes in 1.0.x, the 1.0.x tests would be
skipped -- clearly not what you want.
Also, there should be a way to force the tests regardless of actual
changes; e.g., of you want to restart them after they failed because of
a problem with the testing box, for example.
-- Brane
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org
Received on Sat Jul 24 05:41:52 2004