Mark Phippard wrote:
> Are you talking about the Subclipse tests or svnClientAdapter? I think the
> latter are OK. I have never seen a commit to the Subclipse tests in the
> entire time I have been involved. I assume they do not cover much of
> anything. I do not even know how to run them.
>
I've been tinkering with the former, and no, they haven't been updated
for ages. However, they contain useful tests, and after a bit of
updating, they're failing between 2 and 5 tests, depending on how I run
them, the typical bug being a different svn status than what was expected.
This tells me we may have a race condition with the status cache. Is
that possible? Typically tests have a before and after (like checking if
renaming a resource which was normal before turns it into deleted and
the destination into added). If I set a breakpoint around the action
(move in this case) and step over it, the tests passes. If I let it rip
at full speed, it fails. Mabye there's something async going on which
updates the cache by accident. I'll be looking into this, since I'm
expecting the cache to be transparent. The tests are written before the
cache was added, but that shouldn't matter.
You run them by checking out the plugin, right click and choose Run as
-> Plug-in JUnit test. It starts an Eclipse and does its thing. I'm sure
you can automate it as well, thats how other projects handle their unit
tests (WTP has quite a few)
I'm going to commit my changes one of these days, as pristine it doesn't
even compile.
-Jesper
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subclipse.tigris.org
For additional commands, e-mail: dev-help@subclipse.tigris.org
Received on Fri Oct 20 23:42:56 2006