1254812 - Wrong guess of bug which could have introduced the regression

Status: RESOLVED DUPLICATE of bug 1256379

(Testing :: mozregression, defect)

Description

[Loic]

Run:
mozregression --bits=32 --good=cbe9a2aea9541146ae0a11051ed32d31e
af4e428 --bad=5ddf0a252b086c7e8e468243c6251af5c0b1da57

Result:
 0:10.23 INFO: Last good revision: cbe9a2aea9541146ae0a11051ed32d31eaf4e428
 0:10.23 INFO: First bad revision: 5ddf0a252b086c7e8e468243c6251af5c0b1da57
 0:10.23 INFO: Pushlog:
https://hg.mozilla.org/integration/mozilla-inbound/pushloghtml?fromchange=cbe9a2
aea9541146ae0a11051ed32d31eaf4e428&tochange=5ddf0a252b086c7e8e468243c6251af5c0b1
da57

 0:10.71 INFO: Looks like the following bug has the changes which introduced the
 regression:
https://bugzilla.mozilla.org/show_bug.cgi?id=1175556

But bug 1175556 was a wrong guess, in my case, Bug 1155828 was the faulty.

Comment 1

[Gingerbread Man]

Attachment: mozregression-gui.txt

1. Is the attached log from mozregression-gui 0.9.22 the same as this bug? The log incorrectly pointed to a., which is a backout. The info box in the top right corner b. points to a fairly wide range and therefore isn't that useful. If I manually piece together a pushlog link c., then I get a narrower range.

a. https://hg.mozilla.org/integration/autoland/json-pushes?changeset=d112cf7b2b60e6244099dc3b599a2444ba0d1da3&full=1
b. https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=763909764666b7d617875226e21cce0c8074b959&tochange=479b6858ea6cf96b4842cf5ae78f3bf9e74d3278
c. https://hg.mozilla.org/integration/autoland/pushloghtml?fromchange=6415ccbf739f36f71548540bd92c3b86ab9e1529&tochange=d112cf7b2b60e6244099dc3b599a2444ba0d1da3


2. Is bug 1256379 a duplicate of this? All 3 of these situations involve backouts at least.

Comment 2

[William Lachance (:wlach)]

Yeah, we disabled the automatic bug reporting a while back for exactly this reason.

https://github.com/mozilla/mozregression/blob/44f216c091072b7d6c1c2136041e7c04b0e1c89c/mozregression/main.py#L181

I haven't investigated deeply why this doesn't work and don't feel like it's worth the effort (at least for me) at the moment. I suspect it's the sort of thing where there are so many edge cases that it's extremely difficult to get correct, and I'd rather not even try than risk misleading people (which causes no end of confusion).