994692 - 5.88% dromaeo dom regression on OSX 10.6 from bug 991742

Status: RESOLVED WONTFIX

(Core :: DOM: Core & HTML, defect)

Description

[Joel Maher ( :jmaher ) (UTC -8)]

I found this regression from a talos alert, here is the graph:
http://graphs.mozilla.org/graph.html#tests=[[73,63,21]]&sel=none&displayrange=7&datatype=running

^ note, the first drop is related to bug 984101.

I did some retriggers (we had some coalescing) so I could identify the correct push:
https://tbpl.mozilla.org/?tree=Mozilla-Inbound&fromchange=44732d1a525a&tochange=08aa13c6feac&jobname=Rev4%20MacOSX%20Snow%20Leopard%2010.6%20mozilla-inbound%20talos%20dromaeojs

you can see this in datazilla as well:
https://datazilla.mozilla.org/?start=1396535377&stop=1397140177&product=Firefox&repository=Mozilla-Inbound&os=mac&os_version=OS%20X%2010.8&test=dromaeo_dom&graph_search=2247454df4e5&tr_id=5098366&graph=attr.html&project=talos

we can see that attr.html is regressed.

in bug 990490, we discuss the usefulness of fixing and working on osx 10.6 only regressions.

Comment 1

[Boris Zbarsky [:bzbarsky]]

I don't see any sane way that change would have affected the attr test...  That doesn't even return objects, so doesn't hit the relevant code.

Comment 2

[Benjamin Kerensa [:bkerensa]]

Where did this regress exactly? Also can you explain impact to our users if we chose not to track or block on this.

Comment 3

[Avi Halachmi (:avih)]

(In reply to Benjamin Kerensa [:bkerensa] from comment #2)
> Where did this regress exactly?

If you mean in which tree, then inbound.

> Also can you explain impact to our users if
> we chose not to track or block on this.

Probably no (even if here specifically maybe we can say that JS performance got slightly worse on osx 10.6).

The the new regressions tracking system (bug 990644) is still a new thing and we're not sure of all the details yet. The goal is to flag/file regressions without needing to understand what they mean, but with hopefully enough info to point us at the offending bug.

According to the plan, it's then the job of the offending bug owner to provide more info, hopefully in the form of wontfix (because a/b/c/...) or filing a new bug to fix it (or maybe handle it at the regression bug itself).

Which means that before someone who's familiar with the context makes a comment, we only know about the regression but mostly not of its impact.

Comment 4

[Joel Maher ( :jmaher ) (UTC -8)]

this is limited to osx 10.6 and the regression only affects the dromaeo dom benchmark, specifically the test for attributes.  I assume this is pure javascript related.

Comment 5

[Sylvestre Ledru [:Sylvestre]]

Untracking. Beta 6 should be released today and we won't block the release on this bug.

Comment 6

[u279076]

I'm not sure this bug is valid anymore. I'm not seeing the significant dip we saw back in April evident in the long tail since. I'm also not seeing results since January 13, 2015. Did we stop running this benchmark?

Comment 7

[Joel Maher ( :jmaher ) (UTC -8)]

yes we stopped running this on osx 10.6.  We only run graphics related benchmarks on 10.6- this is too old to be actionable.