Use TimeTicks instead of gettimeofday in ProcessMetrics.

Use TimeTicks instead of gettimeofday in ProcessMetrics.

No real behavior change, but if the user changes their system time there
shouldn't be a jump in %cpu for one call.

BUG=none

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=262059

[Mark Mentovai, 2014-01-23 19:38:00]

https://codereview.chromium.org/139103007/diff/60001/base/process/process_met...
File base/process/process_metrics.h (right):

https://codereview.chromium.org/139103007/diff/60001/base/process/process_met...
base/process/process_metrics.h:208: TimeTicks last_cpu_walltime_;
This (and line 212) aren’t “wall time” if they’re TimeTicks, they’re “ticks.”

[Nico, 2014-01-23 19:40:50]

https://codereview.chromium.org/139103007/diff/60001/base/process/process_met...
File base/process/process_metrics.h (right):

https://codereview.chromium.org/139103007/diff/60001/base/process/process_met...
base/process/process_metrics.h:208: TimeTicks last_cpu_walltime_;
On 2014/01/23 19:38:01, Mark Mentovai wrote:
> This (and line 212) aren’t “wall time” if they’re TimeTicks, they’re “ticks.”

Right, but the linux version already has a last_cpu in "jiffies", and it's
roughly walltime. It's a better name than "cpu time", which this isn't :-)

ticks are ~microseconds in real time, so I think the name isn't so bad?

[Mark Mentovai, 2014-01-23 19:42:30]

The Chrome type for this is TimeTicks. _ticks is the best name.

[Nico, 2014-01-23 19:45:28]

last_cpu_ticks_ doesn't make me think of a time, but of a cpu instruction
count.

*shrug*


On Thu, Jan 23, 2014 at 11:42 AM, <mark@chromium.org> wrote:

> The Chrome type for this is TimeTicks. _ticks is the best name.
>
> https://codereview.chromium.org/139103007/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Mark Mentovai, 2014-01-23 19:47:31]

Then we shouldn’t have named the type TimeTicks.

[Nico, 2014-01-23 19:48:07]

Most of the time it's not used to measure something cpu related :-)


On Thu, Jan 23, 2014 at 11:47 AM, <mark@chromium.org> wrote:

> Then we shouldn’t have named the type TimeTicks.
>
> https://codereview.chromium.org/139103007/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Mark Mentovai, 2014-01-23 19:49:05]

OK, so then:

last_cpu_ticks_
last_cpu_time_ (no change from current name)
last_cpu_, and rename the current Linux last_cpu_ to last_cpu_jiffies_

Any of these are better than last_cpu_walltime_.

[Nico, 2014-01-23 19:52:44]

On Thu, Jan 23, 2014 at 11:49 AM, <mark@chromium.org> wrote:

> OK, so then:
>
> last_cpu_ticks_
> last_cpu_time_ (no change from current name)
> last_cpu_, and rename the current Linux last_cpu_ to last_cpu_jiffies_
>
> Any of these are better than last_cpu_walltime_.
>

Well, I obviously disagree. The first sounds like "cpu ticks", the second
sounds like "cpu time" (which I found actively confusing when looking at
this code today), and the last like "last cpu".


>
> https://codereview.chromium.org/139103007/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Mark Mentovai, 2014-01-23 19:53:54]

t{ime,icks}_at_last_X_measurement_ or something that clarifies that we
don’t really care about counting the ticks?


On Thu, Jan 23, 2014 at 2:49 PM, <mark@chromium.org> wrote:

> OK, so then:
>
> last_cpu_ticks_
> last_cpu_time_ (no change from current name)
> last_cpu_, and rename the current Linux last_cpu_ to last_cpu_jiffies_
>
> Any of these are better than last_cpu_walltime_.
>
> https://codereview.chromium.org/139103007/
>

To unsubscribe from this group and stop receiving emails from it, send an email
to chromium-reviews+unsubscribe@chromium.org.

[Mark Mentovai, 2014-01-23 19:55:04]

Nico wrote:
> Well, I obviously disagree. The first sounds like "cpu ticks", the second
> sounds like "cpu time" (which I found actively confusing when looking at
> this code today), and the last like "last cpu".

My objection is that it’s not a wall time at all. It doesn’t become a wall time
until you take the difference between two and get a TimeDelta.

[Nico, 2014-04-06 00:46:30]

Sorry, wanted to let this CL sit a day and then forgot about it. I kept the name
last_cpu_time_ unchanged now, which we're both not happy with, but that at least
enables this CL to move forward and the renaming can (or can't) happen in
another patch.

[Mark Mentovai, 2014-04-06 01:29:29]

LGTM.

“You're unhappy. I'm unhappy too. Have you heard of Henry Clay? He was the Great
Compromiser. A good compromise is when both parties are dissatisfied, and I
think that's what we have here.” —Larry David

[Nico, 2014-04-06 18:55:12]

The CQ bit was checked by thakis@chromium.org

[commit-bot: I haz the power, 2014-04-06 18:55:14]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/thakis@chromium.org/139103007/220001

[commit-bot: I haz the power, 2014-04-06 21:25:07]

Change committed as 262059