It can be useful to run one or several of the regression tests in a
loop, until one of the tests fail. Here is a patch to do that.

(interesting thing, by the way: you can combine it with -j, in order to
run several instance of the test at once)

+1 I wanted to do this in +1

This is what I wanted to do in http://bugs.python.org/issue7134

I would put this in 2.x and 3.x :)

There was some feedback in this ML thread: 
http://mail.python.org/pipermail/python-dev/2009-June/090238.html

> This is what I wanted to do in http://bugs.python.org/issue7134

Ah, sorry, I hadn't seen it.

Oh, no apologies needed. You have a patch! :)

The patch generally looks good to me, but I think you overlooked line
507, where you do a pending.clear() in case of keyboard interrupt in -j
mode.  I tested it on trunk and it seems to work great, except when I
press ctl-C :)

Thank you, here is an update patch.

I still get:

ValueError: No JSON object could be decoded

in one or more of the threads when I hit ctl-c during a -j run.

I've attached an updated patch that fixes the problem, but I'm not sure
it is a correct fix.

> I've attached an updated patch that fixes the problem, but I'm not sure
> it is a correct fix.

Your patch looks fine to me.

Committed to trunk in r76260 and py3k in r76261.