Add basic daily active use logging.

... will look into some unit and integration testing for all metrics code next in a separate CL.

[sosa, 2010-05-03 20:19:06]

http://codereview.chromium.org/1863002/diff/1/3
File src/platform/metrics/metrics_daemon.cc (right):

http://codereview.chromium.org/1863002/diff/1/3#newcode30
src/platform/metrics/metrics_daemon.cc:30: // The daily use monitor is schedule
to a 1-minute interval after
scheduled*

http://codereview.chromium.org/1863002/diff/1/3#newcode32
src/platform/metrics/metrics_daemon.cc:32: // 10-minute intervals.
Why did you decide to exponentially back off?

http://codereview.chromium.org/1863002/diff/1/3#newcode334
src/platform/metrics/metrics_daemon.cc:334: if (!same_day) {
Else condition is impossible, no?  Use Check instead

http://codereview.chromium.org/1863002/diff/1/3#newcode389
src/platform/metrics/metrics_daemon.cc:389: usemon_source_ =
g_timeout_source_new_seconds(timeout);
I'm a little confused here.  If you already have a timeout source .... have
backoff set to true and using a different timeout ... doesn't this create a
second source with the new timeout without unscheduling the old one?  Do you
then have two sources?  What deletes the other source?

[petkov, 2010-05-03 20:41:33]

Thanks for the quick review. Comments addressed.

http://codereview.chromium.org/1863002/diff/1/3
File src/platform/metrics/metrics_daemon.cc (right):

http://codereview.chromium.org/1863002/diff/1/3#newcode30
src/platform/metrics/metrics_daemon.cc:30: // The daily use monitor is schedule
to a 1-minute interval after
On 2010/05/03 20:19:06, sosa wrote:
> scheduled*
> 

Done.

http://codereview.chromium.org/1863002/diff/1/3#newcode32
src/platform/metrics/metrics_daemon.cc:32: // 10-minute intervals.
On 2010/05/03 20:19:06, sosa wrote:
> Why did you decide to exponentially back off?

I've updated the comments with an explanation. PTAL. Maybe it's an overkill?
It's easy to disable...

http://codereview.chromium.org/1863002/diff/1/3#newcode334
src/platform/metrics/metrics_daemon.cc:334: if (!same_day) {
On 2010/05/03 20:19:06, sosa wrote:
> Else condition is impossible, no?  Use Check instead

The else condition is possible (and actually, the more likely case) -- I've
added a comment clarifying. PTAL.

http://codereview.chromium.org/1863002/diff/1/3#newcode389
src/platform/metrics/metrics_daemon.cc:389: usemon_source_ =
g_timeout_source_new_seconds(timeout);
On 2010/05/03 20:19:06, sosa wrote:
> I'm a little confused here.  If you already have a timeout source .... have
> backoff set to true and using a different timeout ... doesn't this create a
> second source with the new timeout without unscheduling the old one?  Do you
> then have two sources?  What deletes the other source?

GLib will delete the existing source as soon as the current callback returns
FALSE. See the comment in the if (backoff) case above. Should I move the comment
or point to it?

[sosa, 2010-05-03 20:54:18]

lgtm

http://codereview.chromium.org/1863002/diff/1/3
File src/platform/metrics/metrics_daemon.cc (right):

http://codereview.chromium.org/1863002/diff/1/3#newcode389
src/platform/metrics/metrics_daemon.cc:389: usemon_source_ =
g_timeout_source_new_seconds(timeout);
No that's fine.

On 2010/05/03 20:41:33, petkov wrote:
> On 2010/05/03 20:19:06, sosa wrote:
> > I'm a little confused here.  If you already have a timeout source .... have
> > backoff set to true and using a different timeout ... doesn't this create a
> > second source with the new timeout without unscheduling the old one?  Do you
> > then have two sources?  What deletes the other source?
> 
> GLib will delete the existing source as soon as the current callback returns
> FALSE. See the comment in the if (backoff) case above. Should I move the
comment
> or point to it?
> 
> 
>

[petkov, 2010-05-03 21:42:50]

PTAL, added HANDLE_EINTR.

[sosa, 2010-05-03 21:47:12]

lgtm 2x

http://codereview.chromium.org/1863002/diff/6007/12002
File src/platform/metrics/metrics_daemon.cc (right):

http://codereview.chromium.org/1863002/diff/6007/12002#newcode20
src/platform/metrics/metrics_daemon.cc:20: // Jacked from chrome
base/eintr_wrapper.h.
Maybe word choice here? :D

http://codereview.chromium.org/1863002/diff/6007/12002#newcode317
src/platform/metrics/metrics_daemon.cc:317: O_RDWR | O_CREAT, S_IRUSR |
S_IWUSR));
line up arguments

[petkov, 2010-05-03 23:08:54]

Thanks for the review. Pushing...

http://codereview.chromium.org/1863002/diff/6007/12002
File src/platform/metrics/metrics_daemon.cc (right):

http://codereview.chromium.org/1863002/diff/6007/12002#newcode20
src/platform/metrics/metrics_daemon.cc:20: // Jacked from chrome
base/eintr_wrapper.h.
On 2010/05/03 21:47:13, sosa wrote:
> Maybe word choice here? :D

Wording was copied from session_manager_service, actually :-)

I've replaced the macro with an include from base instead so, done.

http://codereview.chromium.org/1863002/diff/6007/12002#newcode317
src/platform/metrics/metrics_daemon.cc:317: O_RDWR | O_CREAT, S_IRUSR |
S_IWUSR));
On 2010/05/03 21:47:13, sosa wrote:
> line up arguments

Done.