Port base/watchdog to Linux.

Port base/watchdog to Linux.

BUG=4632

[please use my chromium address, 2008-11-20 18:55:59]

Tested on Linux, passes unittest.

I'm not sure if my delegate stuff and its communication with Watchdog isn't
totally wrong (in the terms of design). The delegate uses directly its
watchdogs's private fields.

sgk: please note some cleanups to browser_lib.scons. Are they ok?

[Dean McNamee, 2008-11-21 18:26:09]

This is great, thanks.  Looks pretty good, a few comments:

http://codereview.chromium.org/11326/diff/1/5
File base/watchdog.cc (right):

http://codereview.chromium.org/11326/diff/1/5#newcode124
Line 124: std::string watched_name =
WideToASCII(watchdog_->thread_watched_name_);
We're trying to get rid of unneeded wchar_t interfaces, because wide strings /
string literals are 4 bytes per character on Mac and Linux.

What would it take to just make thread_watched_name_ narrow?

http://codereview.chromium.org/11326/diff/1/5#newcode125
Line 125: std::string name = StringPrintf("%s Watchdog", watched_name.c_str());
This could just be concatenation, not a big deal.

http://codereview.chromium.org/11326/diff/1/4
File base/watchdog.h (right):

http://codereview.chromium.org/11326/diff/1/4#newcode52
Line 52: // This is where most of the logic takes place.
This comment does seem very helpful.

[please use my chromium address, 2008-11-21 19:03:53]

http://codereview.chromium.org/11326/diff/1/5
File base/watchdog.cc (right):

http://codereview.chromium.org/11326/diff/1/5#newcode124
Line 124: std::string watched_name =
WideToASCII(watchdog_->thread_watched_name_);
On 2008/11/21 18:26:09, Dean McNamee wrote:
> We're trying to get rid of unneeded wchar_t interfaces, because wide strings /
> string literals are 4 bytes per character on Mac and Linux.
> 
> What would it take to just make thread_watched_name_ narrow?

Done. Tested only on Linux, I wouldn't even bet it compiles on Windows. I will
probably test it on Win over the weekend.

http://codereview.chromium.org/11326/diff/1/5#newcode125
Line 125: std::string name = StringPrintf("%s Watchdog", watched_name.c_str());
On 2008/11/21 18:26:09, Dean McNamee wrote:
> This could just be concatenation, not a big deal.

Done. One #include less.

http://codereview.chromium.org/11326/diff/1/4
File base/watchdog.h (right):

http://codereview.chromium.org/11326/diff/1/4#newcode52
Line 52: // This is where most of the logic takes place.
On 2008/11/21 18:26:09, Dean McNamee wrote:
> This comment does seem very helpful.

Done.

[Dean McNamee, 2008-11-21 19:14:27]

http://codereview.chromium.org/11326/diff/202/20
File chrome/browser/jankometer.cc (right):

http://codereview.chromium.org/11326/diff/202/20#newcode86
Line 86: slow_processing_counter_(std::wstring(L"Chrome.SlowMsg") +
thread_name),
Evan is working on a patch to make all of these counter strings narrow, and then
it will be fine to append them.  Maybe for the time being could you make
JankObserver take a char* ?  Or I guess we can just put a TODO to do it later
when Evan get done with the stats table changes.

http://codereview.chromium.org/11326/diff/202/20#newcode196
Line 196: L"UI",
The string literals are just here, can we make these narrow?

[Dean McNamee, 2008-11-24 07:50:33]

Evan committed his narrow string stats table patch I think.  So now it would
make sense to make all this stuff narrow also?

On 2008/11/21 19:14:27, Dean McNamee wrote:
> http://codereview.chromium.org/11326/diff/202/20
> File chrome/browser/jankometer.cc (right):
> 
> http://codereview.chromium.org/11326/diff/202/20#newcode86
> Line 86: slow_processing_counter_(std::wstring(L"Chrome.SlowMsg") +
> thread_name),
> Evan is working on a patch to make all of these counter strings narrow, and
then
> it will be fine to append them.  Maybe for the time being could you make
> JankObserver take a char* ?  Or I guess we can just put a TODO to do it later
> when Evan get done with the stats table changes.
> 
> http://codereview.chromium.org/11326/diff/202/20#newcode196
> Line 196: L"UI",
> The string literals are just here, can we make these narrow?

[please use my chromium address, 2008-11-24 11:42:32]

Sure. Please take a look now.

[Dean McNamee, 2008-11-24 16:25:35]

Looks great, thanks again.  I will have Jar do a quick peek and try committing
it today.

[Dean McNamee, 2008-11-25 21:30:34]

On 2008/11/24 16:25:35, Dean McNamee wrote:
> Looks great, thanks again.  I will have Jar do a quick peek and try committing
> it today.

No reply from Jar.  Sent to the try servers and will commit if all is ok.

[Dean McNamee, 2008-11-25 22:48:09]

Committed as r6004.

On 2008/11/25 21:30:34, Dean McNamee wrote:
> On 2008/11/24 16:25:35, Dean McNamee wrote:
> > Looks great, thanks again.  I will have Jar do a quick peek and try
committing
> > it today.
> 
> No reply from Jar.  Sent to the try servers and will commit if all is ok.

[wtc, 2008-11-25 23:00:13]

LGTM.

I reviewed this CL for jar because he's busy today.

http://codereview.chromium.org/11326/diff/45/215
File base/base_unittests.scons (right):

http://codereview.chromium.org/11326/diff/45/215#newcode50
Line 50: # These test files work on *all* platforms; tests that don't work
Should we also update this comment as we updated the same
comment in base/base_lib.scons?

http://codereview.chromium.org/11326/diff/45/217
File base/watchdog.h (right):

http://codereview.chromium.org/11326/diff/45/217#newcode53
Line 53: explicit ThreadDelegate(Watchdog* watchdog, const base::TimeDelta&
duration)
This constructor has two parameters, so it doesn't need to
be 'explicit'.