Restart the test server for every test to clear the db, to reduce the likehood

Add easy way to make it verbose and enable it until tests are fixed.

Fix django version since rietveld expects 1.0.x.

[Evan Martin, 2010-08-25 17:06:05]

http://codereview.chromium.org/3165060/diff/1/2
File PRESUBMIT.py (right):

http://codereview.chromium.org/3165060/diff/1/2#newcode68
PRESUBMIT.py:68: 'svn/django/branches/releases/1.0.X/django',
Would it be better to pick a specific rev from this dir?  I worry different
people will have different checkout versions causing different test behaviors.

http://codereview.chromium.org/3165060/diff/1/2#newcode85
PRESUBMIT.py:85: continue
We probably should move this logic to a 'run-all-tests.sh' or something
eventually.

http://codereview.chromium.org/3165060/diff/1/2#newcode94
PRESUBMIT.py:94: # Restart the dev server for every test to clear the DB.
Why is this needed?  Doesn't it make it slower?

[M-A Ruel, 2010-08-25 19:24:36]

http://codereview.chromium.org/3165060/diff/1/2
File PRESUBMIT.py (right):

http://codereview.chromium.org/3165060/diff/1/2#newcode85
PRESUBMIT.py:85: continue
On 2010/08/25 17:06:05, Evan Martin wrote:
> We probably should move this logic to a 'run-all-tests.sh' or something
> eventually.

Yeah? Why wasn't it there already? :P

http://codereview.chromium.org/3165060/diff/1/2#newcode94
PRESUBMIT.py:94: # Restart the dev server for every test to clear the DB.
On 2010/08/25 17:06:05, Evan Martin wrote:
> Why is this needed?  Doesn't it make it slower?

Yes it's slower but it's already slow as molasses since the scripts are
recreating the svn and git repos on each test.

I'm just trying to pinpoint what's broken.

Right now, rietveld barfs because it seems my 1.3.5 dev server doesn't support
db.UserProperty(auto_current_user_add=True, required=True) correctly.

[Evan Martin, 2010-08-25 19:30:30]

On 2010/08/25 19:24:36, Marc-Antoine Ruel wrote:
> Yes it's slower but it's already slow as molasses since the scripts are
> recreating the svn and git repos on each test.
> 
> I'm just trying to pinpoint what's broken.
> 
> Right now, rietveld barfs because it seems my 1.3.5 dev server doesn't support
> db.UserProperty(auto_current_user_add=True, required=True) correctly.

So you're saying that the tests don't pass due to some bug, right?  I guess I
don't see why you're making the presubmit slower when there's a bug in the tests
-- the tests are conceptually failing already, in that they can't finish without
recreating the server.

[M-A Ruel, 2010-08-25 19:37:11]

On 2010/08/25 19:30:30, Evan Martin wrote:
> On 2010/08/25 19:24:36, Marc-Antoine Ruel wrote:
> > Yes it's slower but it's already slow as molasses since the scripts are
> > recreating the svn and git repos on each test.
> > 
> > I'm just trying to pinpoint what's broken.
> > 
> > Right now, rietveld barfs because it seems my 1.3.5 dev server doesn't
support
> > db.UserProperty(auto_current_user_add=True, required=True) correctly.
> 
> So you're saying that the tests don't pass due to some bug, right?  I guess I
> don't see why you're making the presubmit slower when there's a bug in the
tests
> -- the tests are conceptually failing already, in that they can't finish
without
> recreating the server.

Done and uploaded.

Did you ever run the test and having them pass? Which rietveld version were you
using?

[Evan Martin, 2010-08-25 19:41:39]

On 2010/08/25 19:37:11, Marc-Antoine Ruel wrote:
> Did you ever run the test and having them pass? Which rietveld version were
you
> using?

It's been a while since I made a change or ran the tests.  :\

[Evan Martin, 2010-08-25 19:42:01]

Can you update the review description now?

[M-A Ruel, 2010-08-25 19:46:08]

On 2010/08/25 19:42:01, Evan Martin wrote:
> Can you update the review description now?

oops, done.

[Evan Martin, 2010-08-25 20:09:03]

LGTM