content: Terminate early if the toolkit initialization fails.

content: Terminate early if the toolkit initialization fails.

On Chrome OS, it is possible for chrome to start without the X11
server (e.g. X server has crashed, and hasn't restarted since). In
such cases, Chrome ends up causing a crash. So instead of the crash,
abort the startup sequence if the toolkit initialization step fails.

BUG=364929
R=jam@chromium.org

Committed: https://src.chromium.org/viewvc/chrome?view=rev&revision=266231

[sadrul, 2014-04-24 22:28:57]

Hi! Does this look like a reasonable way to handle failures during
initialization?

[Daniel Erat, 2014-04-24 22:39:27]

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
content/browser/browser_main_runner.cc:88: return 0;
this method is super-undocumented. is this used as the program's exit code on
unix? if so, you should probably return a positive number instead of 0 -- chrome
_did_ fail to start.

[sadrul, 2014-04-24 22:57:13]

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
content/browser/browser_main_runner.cc:88: return 0;
On 2014/04/24 22:39:27, Daniel Erat wrote:
> this method is super-undocumented. is this used as the program's exit code on
> unix?

Yep, looks like it.

> if so, you should probably return a positive number instead of 0 -- chrome
> _did_ fail to start.

Is there a convention for what the exit-code should be in this case? Or should I
just use any random positive number? In that case, any suggestion/preference?

[Daniel Erat, 2014-04-24 23:21:07]

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
content/browser/browser_main_runner.cc:88: return 0;
On 2014/04/24 22:57:14, sadrul wrote:
> On 2014/04/24 22:39:27, Daniel Erat wrote:
> > this method is super-undocumented. is this used as the program's exit code
on
> > unix?
> 
> Yep, looks like it.
> 
> > if so, you should probably return a positive number instead of 0 -- chrome
> > _did_ fail to start.
> 
> Is there a convention for what the exit-code should be in this case? Or should
I
> just use any random positive number? In that case, any suggestion/preference?

there aren't any conventions, as far as i'm aware. it's pretty common to use 1
as a generic code for failure, so i'd recommend that here.

[sadrul, 2014-04-25 02:34:47]

I also added a comment for BrowserMainRunner::Initialize() to mention the
significance of the returned value.

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/1/content/browser/browser_main...
content/browser/browser_main_runner.cc:88: return 0;
On 2014/04/24 23:21:08, Daniel Erat wrote:
> On 2014/04/24 22:57:14, sadrul wrote:
> > On 2014/04/24 22:39:27, Daniel Erat wrote:
> > > this method is super-undocumented. is this used as the program's exit code
> on
> > > unix?
> > 
> > Yep, looks like it.
> > 
> > > if so, you should probably return a positive number instead of 0 -- chrome
> > > _did_ fail to start.
> > 
> > Is there a convention for what the exit-code should be in this case? Or
should
> I
> > just use any random positive number? In that case, any
suggestion/preference?
> 
> there aren't any conventions, as far as i'm aware. it's pretty common to use 1
> as a generic code for failure, so i'd recommend that here.

Done.

[jam, 2014-04-25 05:05:53]

https://codereview.chromium.org/250723002/diff/20001/content/browser/browser_...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/20001/content/browser/browser_...
content/browser/browser_main_runner.cc:88: return 1;
the comment says 0 or negative means initialization failed, but you're returning
1?

[sadrul, 2014-04-25 10:43:35]

https://codereview.chromium.org/250723002/diff/20001/content/browser/browser_...
File content/browser/browser_main_runner.cc (right):

https://codereview.chromium.org/250723002/diff/20001/content/browser/browser_...
content/browser/browser_main_runner.cc:88: return 1;
On 2014/04/25 05:05:53, jam wrote:
> the comment says 0 or negative means initialization failed, but you're
returning
> 1?

I messed up the comment (I think I started with 'zero or positive' and decided
halfway through that 'non-negative' would be better, and just messed up the
comment). Whoops. Fixed.

[jam, 2014-04-25 18:43:23]

lgtm

[sadrul, 2014-04-25 20:17:23]

Committed patchset #3 manually as r266231 (presubmit successful).