linux: expose the ProcessSingleton timeout to speed tests

linux: expose the ProcessSingleton timeout to speed tests

We have a 20-second timeout normally, but for testing purposes 1 second is
plenty.

[Paweł Hajdan Jr., 2009-09-17 19:45:56]

After fixing the nits, LGTM. Thanks!

http://codereview.chromium.org/209018/diff/1/3
File chrome/browser/process_singleton_linux.cc (right):

http://codereview.chromium.org/209018/diff/1/3#newcode630
Line 630: return NotifyOtherProcessWithTimeout(kTimeoutInSeconds /* seconds */);
nit: Is this /* seconds */ comment necessary? The constant itself ends with
Seconds...

http://codereview.chromium.org/209018/diff/1/4
File chrome/browser/process_singleton_linux_uitest.cc (right):

http://codereview.chromium.org/209018/diff/1/4#newcode62
Line 62: const int kTimeoutSeconds = 1;
I'd be a bit more conservative and put 3 instead of 1 here.

Where we get to a point when even that is too slow, that's good. But for now all
50 slowest tests have run time over 5 seconds.

[Evan Stade, 2009-09-17 20:15:58]

I think maybe this will fix bug 22098.

[Evan Martin, 2009-09-17 20:37:13]

http://codereview.chromium.org/209018/diff/1/3
File chrome/browser/process_singleton_linux.cc (right):

http://codereview.chromium.org/209018/diff/1/3#newcode630
Line 630: return NotifyOtherProcessWithTimeout(kTimeoutInSeconds /* seconds */);
On 2009/09/17 19:45:56, Paweł Hajdan Jr. wrote:
> nit: Is this /* seconds */ comment necessary? The constant itself ends with
> Seconds...

Good catch.  It was leftover from before I had noticed kTimeoutInSeconds existed
(you can see in the code below it used to have '20' hardcoded).

[dank, 2009-09-17 21:17:33]

Just curious, will this behave nicely under valgrind, where things is slow?

[Evan Martin, 2009-09-17 21:21:43]

On 2009/09/17 21:17:33, dank wrote:
> Just curious, will this behave nicely under valgrind, where things is slow?

I upped the timeout to 3 seconds on Pawel's request.  I guess we'll find out...
:\