> On Feb 17, 2020, at 5:04 AM, Pierre Smits <pierresm...@apache.org> wrote:
> It seems to me that when credentials are provided, the script does a
> authentication first before test are executed, and then something starts to
> go wrong.
>
> Or am I doing something wrong?
A troubleshooting tip is to look at the patch-dryrun.log in the
--patch-dir. That will give you some hints as to why the patch couldn't be
applied.
That said, given your previous command lines, I have a suspicion that
the repo isn't clean. --dirty-workspace, in particular, will force test-patch
not to try and remove remnants of things. So the patch has already been applied
and therefore can't be applied again/leads to general confusion as test-patch
will pick up other things that have changed outside of the patch (since a lot
of the work has to be done post-compilation).
In general, I personally recommend that when using test-patch with PRs,
create a fresh repo outside of your normal development repo, set --basedir to
your new testing repo, and add the --resetrepo flag to test-patch. That last
flag will make sure that test-patch has a fresh working slate to apply things
by forcibly cleaning it out, doing pulls, setting the branch to
trunk/master/whatever, etc.
