Send client ID in crash report

Send client ID in crash report

BUG=4110, 988

[petkov, 2010-07-01 06:33:45]

A few nits/questions, LGTM otherwise.

http://codereview.chromium.org/2849038/diff/1/3
File crash_sender (right):

http://codereview.chromium.org/2849038/diff/1/3#newcode20
crash_sender:20: MINIDUMP_UPLOADER=/usr/bin/minidump_upload
Unused? Remove if so...

http://codereview.chromium.org/2849038/diff/1/3#newcode39
crash_sender:39: SECONDS_SEND_SPREAD=${SECONDS_SEND_SPREADS:-600}
I assume you've tested this with dash?

http://codereview.chromium.org/2849038/diff/1/3#newcode88
crash_sender:88: # See crosbug.com/3303.
Do you still need to issue reference? If so, it should probably point directly
to 988 instead.

http://codereview.chromium.org/2849038/diff/1/3#newcode89
crash_sender:89: if [ -f "${CONSENT_ID}" ]; then
Maybe -r instead of -f ?

Also, it almost hurts when I see 5 lines instead of 2 but, again, no style
requirement :-)

http://codereview.chromium.org/2849038/diff/1/3#newcode167
crash_sender:167: lecho "Crash report receipt ID $(cat ${report_id})"
It might make sense to include recent receipt IDs in about:system. Or do you
assume users can get this info from syslog?

[kmixter1, 2010-07-01 21:33:37]

pushing...

http://codereview.chromium.org/2849038/diff/1/3
File crash_sender (right):

http://codereview.chromium.org/2849038/diff/1/3#newcode20
crash_sender:20: MINIDUMP_UPLOADER=/usr/bin/minidump_upload
On 2010/07/01 06:33:45, petkov wrote:
> Unused? Remove if so...
> 

Done.

http://codereview.chromium.org/2849038/diff/1/3#newcode39
crash_sender:39: SECONDS_SEND_SPREAD=${SECONDS_SEND_SPREADS:-600}
On 2010/07/01 06:33:45, petkov wrote:
> I assume you've tested this with dash?
> 

The autotest did.  Also, I've used the external override to send these
immediately - though you did point out a typo.

http://codereview.chromium.org/2849038/diff/1/3#newcode88
crash_sender:88: # See crosbug.com/3303.
On 2010/07/01 06:33:45, petkov wrote:
> Do you still need to issue reference? If so, it should probably point directly
> to 988 instead.

Done.

http://codereview.chromium.org/2849038/diff/1/3#newcode89
crash_sender:89: if [ -f "${CONSENT_ID}" ]; then
On 2010/07/01 06:33:45, petkov wrote:
> Maybe -r instead of -f ?
> 
> Also, it almost hurts when I see 5 lines instead of 2 but, again, no style
> requirement :-)
> 

I generally like this if/else formatting better since it's the same number of
lines as in C (though admittedly in this specific case you could return the
value immediately instead of branching on it).  But I've changed it to be less
painful to you. ;)

http://codereview.chromium.org/2849038/diff/1/3#newcode167
crash_sender:167: lecho "Crash report receipt ID $(cat ${report_id})"
On 2010/07/01 06:33:45, petkov wrote:
> It might make sense to include recent receipt IDs in about:system. Or do you
> assume users can get this info from syslog?
> 

From syslog - this is only useful to developers and they will know about syslogs
which also show up in about:system.