Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss

Talos Regression: Ts, Cold increase 12.04% on Leopard Firefox

0 views
Skip to first unread message

Nobody

unread,
Nov 13, 2009, 1:16:01 AM11/13/09
to dev-tree-...@lists.mozilla.org

Regression: Ts, Cold increase 12.04% on Leopard Firefox

Previous results: 12124.2 from build 20091112141427 of revision 3dae66874786 at 2009-11-12 15:07:00 on talos-rev2-leopard02 run # 0

New results: 13583.4 from build 20091112145835 of revision e3fa67bf4fa0 at 2009-11-12 15:58:00 on talos-rev2-leopard15 run # 0

Suspected checkin range: from revision 3dae66874786 to revision e3fa67bf4fa0

Mike Beltzner

unread,
Nov 13, 2009, 10:28:01 AM11/13/09
to Alice Nodelman, dev-tree-...@lists.mozilla.org
The regression range is: https://bugzilla.mozilla.org/show_bug.cgi?id=489586 which is making the UA for 64 bit Firefox "MacIntel" - no way, really, that it was caused by the code change.

Machine issue?

_______________________________________________
dev-tree-management mailing list
dev-tree-...@lists.mozilla.org
https://lists.mozilla.org/listinfo/dev-tree-management

Johnathan Nightingale

unread,
Nov 13, 2009, 10:32:14 AM11/13/09
to Mike Beltzner, dev-tree-...@lists.mozilla.org
And the graph link doesn't show the bump, to my eyes. ( http://people.mozilla.org/~johnath/sattap/8db8c786.png )

Catlee - any thoughts on how to make us more noise-resistant here? I don't want to start having false negatives, I think you've made exactly the right decision to risk some false-positives (identified as regressions where a human wouldn't) in the name of catching every false negative (failing to send email when a human would have). I guess I'm just wondering if you can be even MORE awesome, because that is how I reward people who do awesome things, because I am terrible.

J

---
Johnathan Nightingale
Human Shield

Chris AtLee

unread,
Nov 13, 2009, 10:53:21 AM11/13/09
to Johnathan Nightingale, dev-tree-...@lists.mozilla.org, Mike Beltzner
Well, I see what the script is picking up on:
http://people.mozilla.org/~catlee/sattap/f5e717e6.png

There are a sequence of 5 or 6 data points that are high. The first one
was probably within the noise, so it ignored it and warned about the second.

The first data point that looks high to me is
http://hg.mozilla.org/mozilla-central/rev/8c387f11fda1

However, it looks like it's come down again since then? Maybe we got
unlucky and reported on the high end of the noise a few times in a row.
The only way I can think of to reduce these false positives, is to
increase the number of data points we wait for before making a
determination, which will delay reporting of real regressions.

BTW, the graph server doesn't always show the correct revision. Yay.
This is https://bugzilla.mozilla.org/show_bug.cgi?id=500194

On 13/11/09 10:32 AM, Johnathan Nightingale wrote:
> And the graph link doesn't show the bump, to my eyes. (
> http://people.mozilla.org/~johnath/sattap/8db8c786.png )
>
> Catlee - any thoughts on how to make us more noise-resistant here? I
> don't want to start having false negatives, I think you've made exactly
> the right decision to risk some false-positives (identified as
> regressions where a human wouldn't) in the name of catching every false
> negative (failing to send email when a human would have). I guess I'm
> just wondering if you can be even MORE awesome, because that is how I
> reward people who do awesome things, because I am terrible.
>
> J
>
> On 13-Nov-09, at 10:28 AM, Mike Beltzner wrote:
>
>> The regression range is:
>> https://bugzilla.mozilla.org/show_bug.cgi?id=489586 which is making
>> the UA for 64 bit Firefox "MacIntel" - no way, really, that it was
>> caused by the code change.
>>
>> Machine issue?
>>
>> On 2009-11-13, at 1:16 AM, Nobody wrote:
>>
>>> Regression: Ts, Cold increase 12.04%

>>> <http://graphs.mozilla.org/graph.html#tests=[{%22machine%22:169,%22test%22:55,%22branch%22:1},{%22machine%22:170,%22test%22:55,%22branch%22:1},{%22machine%22:172,%22test%22:55,%22branch%22:1},{%22machine%22:173,%22test%22:55,%22branch%22:1},{%22machine%22:174,%22test%22:55,%22branch%22:1},{%22machine%22:175,%22test%22:55,%22branch%22:1},{%22machine%22:176,%22test%22:55,%22branch%22:1},{%22machine%22:177,%22test%22:55,%22branch%22:1},{%22machine%22:178,%22test%22:55,%22branch%22:1},{%22machine%22:179,%22test%22:55,%22branch%22:1},{%22machine%22:180,%22test%22:55,%22branch%22:1},{%22machine%22:181,%22test%22:55,%22branch%22:1},{%22machine%22:182,%22test%22:55,%22branch%22:1}]&sel=1257983880,1258156680>


>>> on Leopard Firefox
>>>
>>> Previous results: 12124.2 from build 20091112141427 of revision
>>> 3dae66874786 at 2009-11-12 15:07:00 on talos-rev2-leopard02 run # 0
>>>
>>> New results: 13583.4 from build 20091112145835 of revision
>>> e3fa67bf4fa0 at 2009-11-12 15:58:00 on talos-rev2-leopard15 run # 0
>>>
>>> Suspected checkin range: from revision 3dae66874786 to revision
>>> e3fa67bf4fa0

>>> <x-msg://170/%20%20%20%20http://hg.mozilla.org/mozilla-central/pushloghtml?fromchange=3dae66874786&tochange=e3fa67bf4fa0>


>>>
>>> _______________________________________________
>>> dev-tree-management mailing list
>>> dev-tree-...@lists.mozilla.org

>>> <mailto:dev-tree-...@lists.mozilla.org>


>>> https://lists.mozilla.org/listinfo/dev-tree-management
>>
>> _______________________________________________
>> dev-tree-management mailing list
>> dev-tree-...@lists.mozilla.org

>> <mailto:dev-tree-...@lists.mozilla.org>


>> https://lists.mozilla.org/listinfo/dev-tree-management
>
> ---
> Johnathan Nightingale
> Human Shield

> joh...@mozilla.com <mailto:joh...@mozilla.com>

0 new messages