talkback lookup now available
Asa Dotzler
look-up tool.
The new tool is located at http://talkback-public.mozilla.org. From this
page you can look up individual talkback reports . The reports contain
the stack signature, product and build IDs, the time of the crash, the
platform and OS, the code module, the user-entered URL and user
comments, the trigger reason, trigger line number, and stack trace for
the crash.
The stack lines in the stack trace are each linked to a bugzilla query
for existing bug reports with that class and function in the comments.
This makes it easy to discover if your crasher is already reported. Just
look up your talkback ID, click the top line of the stack trace and have
a look at the resulting bug list to see if one of those reports matches
your crasher.
In addition to the individual incident ID look-ups, you can also view
our talkback "smart analysis" reports from that same page. These reports
give some aggregate information and include our current list of topcrashes.
With the good news of this long-awaited triage tool comes some bad news
that we lost a large number of previous crash reports because of some
database problems we were having. So if you go to look up your incident
from a week or two ago and don't get a report, that's the reason. Try
crashing again and get a newer incident :-)
--Asa
Boris Zbarsky
> Thanks to Jay's hard work, we now have a publically available talkback
> look-up tool.
Asa, that's wonderful news!
If someone can mail me (personal mail is fine) a talkback ID for a crash
(anything that actually brings up info when using this tool), that would
be much appreciated.
> In addition to the individual incident ID look-ups, you can also view
> our talkback "smart analysis" reports from that same page. These reports
> give some aggregate information and include our current list of topcrashes.
Hmm... Is this part not up yet?
-Boris
Asa Dotzler
> Asa Dotzler wrote:
>
>> Thanks to Jay's hard work, we now have a publically available talkback
>> look-up tool.
>
>
> Asa, that's wonderful news!
>
> If someone can mail me (personal mail is fine) a talkback ID for a crash
> (anything that actually brings up info when using this tool), that would
> be much appreciated.
Try 13843 (I just picked a random recent report)
>
>> In addition to the individual incident ID look-ups, you can also view
>> our talkback "smart analysis" reports from that same page. These
>> reports give some aggregate information and include our current list
>> of topcrashes.
>
>
> Hmm... Is this part not up yet?
Yeah, but it attempts to do it in a pop-up window and seems to get
caught by our pop-up blocking or something. I'll talk to Jay about this.
It should use the same window. You can either whitelist it or just
visithttp://talkback-public.mozilla.org/reports/ for the various reports.
--Asa
Daniel Veditz
> Asa Dotzler wrote:
>
>>our talkback "smart analysis" reports from that same page. These reports
>>give some aggregate information and include our current list of topcrashes.
>
> Hmm... Is this part not up yet?
The UI is not intuitive, and it's foiled by the popup-blocker. When you
change the right-hand drop-down an onchange() handler tries to open a new
window.
A better UI would be to add a "go" button to that control as well to give
people more of a hint what to do, and a click on a button would avoid the
popup blocker. In the meantime add talkback-public.mozilla.org to the popup
whitelist.
-Dan Veditz
Boris Zbarsky
> Try 13843 (I just picked a random recent report)
Ah, ok. So this just does stack traces for now, no local vars, right?
Is that planned, or are there privacy issues?
-Boris
L. David Baron
Local variables / raw stack could have privacy issues. Disassembly
(i.e., exactly which instruction caused the crash) and registers don't
(as far as I know), and are often quite useful as well. And if it's
needed, people at mozilla.org should be able to get the raw stack data.
-David
--
L. David Baron <URL: http://dbaron.org/ >
Boris Zbarsky
> Local variables / raw stack could have privacy issues. Disassembly
> (i.e., exactly which instruction caused the crash) and registers don't
> (as far as I know), and are often quite useful as well.
Useful, yes. But require a lot more expertise to interpret than the
local variable values....
Perhaps we could output information for each local var along the lines
of "is it 0?"
> And if it's needed, people at mozilla.org should be able to get the raw stack data.
Fair enough. Would it be possible to set up some sort of semi-formal
process for requesting such (a request flag or an email address to cc)?
Or do we not foresee this coming up much and want to handle it on an
ad-hoc basis for now?
-Boris
Asa Dotzler
We can do any of the above or we can try to give more complete access to
a few more trusted people. In the near term, an email to me or anyone
else inside the MF firewall would probably yield results.
I recently created the "talkbackid" keyword for people to request stack
traces from talkback and now that it's mooted, maybe we could replace it
with a keyword for requesting local variable values -- though I
personally don't know enough about interpreting that data to make sure I
wouldn't disclose privacy-sensitive data. I'd probably turn to someone
like dbaron for help.
--Asa
Riccardo Cretti
> Thanks to Jay's hard work, we now have a publically available talkback
> look-up tool.
Wow, that's great!
Now if only I could make my browser crash more!!! :-P
--
*Riccardo Cretti*
squad...@jumpy.it - ICQ 15927849
A proud owner of 2 Amiga, a PC and a C64! :-)
Reality is for people who can't handle Star Trek
James Ross
> Thanks to Jay's hard work, we now have a publically available talkback
> look-up tool.
>
> The new tool is located at http://talkback-public.mozilla.org. From this
> page you can look up individual talkback reports . The reports contain
> the stack signature, product and build IDs, the time of the crash, the
> platform and OS, the code module, the user-entered URL and user
> comments, the trigger reason, trigger line number, and stack trace for
> the crash.
Sweet.
But, I have a question. Is there some time-delay between crashing,
getting a Talkback ID, and being able to find it with this?
Currently the latest TBID I can see (without getting the message saying
it doesn't exist) is 15057 (dated 8th), but my last crash has TBID 16032
(dated 10th).
--
James Ross <sil...@warwickcompsoc.co.uk>
Terribly discriminatory isn't it, young man? It seems that Yusaku
displays kindness only to half-naked women.
-- Mr. Yotsuya
(Maison Ikkoku, by Rumiko Takahashi)
Asa Dotzler
> Asa Dotzler wrote:
>
>> Thanks to Jay's hard work, we now have a publically available talkback
>> look-up tool.
>>
>> The new tool is located at http://talkback-public.mozilla.org. From
>> this page you can look up individual talkback reports . The reports
>> contain the stack signature, product and build IDs, the time of the
>> crash, the platform and OS, the code module, the user-entered URL and
>> user comments, the trigger reason, trigger line number, and stack
>> trace for the crash.
>
>
> Sweet.
>
> But, I have a question. Is there some time-delay between crashing,
> getting a Talkback ID, and being able to find it with this?
>
> Currently the latest TBID I can see (without getting the message saying
> it doesn't exist) is 15057 (dated 8th), but my last crash has TBID 16032
> (dated 10th).
>
Yeah, it seems we're having some kind of congestion in the system. I
just attempted to look up a recent crash I submitted and it was behind
about 1800 reports. I've informed the right people and hopefully they'll
be able to make things better.
--Asa