[ANN] TortoiseHg-0.3

[Steve Borho]

On Sun, 2008-02-03 at 11:13 -0800, David Ripton wrote:
> On 2008.02.02 22:54:06 -0600, Steve Borho wrote:
> > On Sat, 2008-02-02 at 20:07 -0800, Brad Schick wrote:
> > > -Would be nice to have a way to cancel running datamining searches. I 
> > > ran two regexp searches that consumed 100% of the cpu and didn't end 
> > > until I killed python (which I now can't repro).
> > 
> > Anyone know how to stop a Python thread?
> 
> There's no built-in, safe, easy way.
> 
> You can write your threads such that they suicide upon command:
> 
> class KillableThread(threading.Thread):
>     time_to_die = False
> 
>     def please_die(self):
>         self.time_to_die = True
> 
>     def run(self):
>         while True:
>             if self.time_to_die:
>                 return
>             # Do a small amount of work here, without blocking too long

Most of TortoiseHg threads are calling the mercurial dispatch function,
so this usually isn't an option for us.  Though I did use this method
for the web server dialog.

> But if you're calling a blocking library without timeouts that's outside
> your control, then that doesn't really work.
> 
> One (cool, but possibly brittle or non-portable) option is to mess
> around inside Python's threading internals like:
> http://sebulba.wikispaces.com/recipe+thread2

Now that looks interesting.  Thanks for the link.

> Another option is to call such libraries from within a subprocess.
> Killing processes is portable.

This would be too heavy weight for most operations we do.

-- 
Steve Borho (steve at borho.org)
http://www.borho.org/~steve/steve.asc
Key fingerprint = 2D08 E7CF B624 624C DE1F  E2E4 B0C2 5292 F2C6 2C8C