Convert chrome/common non-debug logs to debug logs.

Convert chrome/common non-debug logs to debug logs.
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=107814

[jam, 2011-10-28 20:40:01]

lgtm

[Scott Hess - ex-Googler, 2011-10-28 23:46:50]

Hey ... a number of these in the Mac code were intentionally CHECK() because if
they really are wrong in production, we really don't want to continue, and
really do want a crash, so that we'll be able to resolve it.  I realize you
don't want to embark on a centithread on this, but I do think that the closer
you get to platform-support code the more likely it is that you're changing
intended behavior rather than accidental stuff.

[brettw, 2011-10-29 02:19:11]

On Fri, Oct 28, 2011 at 4:46 PM,  <shess@chromium.org> wrote:
> Hey ... a number of these in the Mac code were intentionally CHECK() because
> if
> they really are wrong in production, we really don't want to continue, and
> really do want a crash, so that we'll be able to resolve it.  I realize you
> don't want to embark on a centithread on this, but I do think that the
> closer
> you get to platform-support code the more likely it is that you're changing
> intended behavior rather than accidental stuff.

Sure, which ones should I put back?

Brett

[Scott Hess - ex-Googler, 2011-10-29 20:27:31]

I'll circle back on Monday and get them, then I can post them to
someone who can appropriately challenge whether they're really
necessary.  It's more that I only happened across this quite by
chance, and am a tad worried that there will be other interesting
occurrences which won't be noticed until later.  It might be
reasonable to have an associated bug or something to centrally track
broad semi-automated changes of this sort.

I'm not so concerned about the LOG->DLOG as I am about the
FATAL->debug_FATAL changes.  Losing all-the-time logging is sad, but
just means that it takes longer to debug something.  Actual
functioning of Chrome does not change.  Converting CHECK to DCHECK
means that we might be exposing entirely new code paths in production
which were previously impossible, intentionally so.

-scott


On Fri, Oct 28, 2011 at 7:19 PM, Brett Wilson <brettw@chromium.org> wrote:
> On Fri, Oct 28, 2011 at 4:46 PM,  <shess@chromium.org> wrote:
>> Hey ... a number of these in the Mac code were intentionally CHECK() because
>> if
>> they really are wrong in production, we really don't want to continue, and
>> really do want a crash, so that we'll be able to resolve it.  I realize you
>> don't want to embark on a centithread on this, but I do think that the
>> closer
>> you get to platform-support code the more likely it is that you're changing
>> intended behavior rather than accidental stuff.
>
> Sure, which ones should I put back?
>
> Brett
>

[brettw, 2011-10-29 23:11:55]

On Sat, Oct 29, 2011 at 1:27 PM, Scott Hess <shess@chromium.org> wrote:
> I'll circle back on Monday and get them, then I can post them to
> someone who can appropriately challenge whether they're really
> necessary.  It's more that I only happened across this quite by
> chance, and am a tad worried that there will be other interesting
> occurrences which won't be noticed until later.  It might be
> reasonable to have an associated bug or something to centrally track
> broad semi-automated changes of this sort.
>
> I'm not so concerned about the LOG->DLOG as I am about the
> FATAL->debug_FATAL changes.  Losing all-the-time logging is sad, but
> just means that it takes longer to debug something.  Actual
> functioning of Chrome does not change.  Converting CHECK to DCHECK
> means that we might be exposing entirely new code paths in production
> which were previously impossible, intentionally so.

I didn't change most of the CHECKs, only the ones that seemed less
critical. It's possible I shouldn't have changed a few of the Mac ones
on account of not understanding some history.

Brett

[Aaron Boodman, 2011-11-05 16:00:39]

On 2011/10/29 23:11:55, brettw wrote:
> I didn't change most of the CHECKs, only the ones that seemed less
> critical.

I would also prefer having all the DCHECKs you changed in extensions and
JSONSchemaValidator put back to CHECK for reasons I've already argued. I also
really don't like things being inconsistent within one file.

[brettw, 2011-11-05 16:49:39]

Sorry I can't do any coding now. Please feel free to TBR partial
reverts you think were inappropriate.

Brett