1005920 - The prevent_webidl_changes hook must ignore backouts

Status: RESOLVED FIXED

(Developer Services :: Mercurial: hg.mozilla.org, defect)

Description

[(no longer active)]

Do we have good criteria for detecting backouts?  Or should I just look for "backout" and "backed out" in the commit message?

Comment 1

[Ryan VanderMeulen [:RyanVM]]

"revert" would probably be good too

Comment 2

[(no longer active)]

Attachment: Patch (v1)

Comment 3

[(no longer active)]

http://hg.mozilla.org/hgcustom/hghooks/rev/94e6623eb576

Comment 4

[Justin Wood (:Callek)]

ehsan sudden thought, wouldn't a backout of a cset that landed on trunk affecting .webidl be equally as painful for web compat and stability as a .webidl changing due to new functionality?

As in, is this "really" the right thing here. (I'm just drawing attention to a case that I didn't see called out, if you already thought of it and this is still desired, great.)

Comment 5

[(no longer active)]

(In reply to Justin Wood (:Callek) from comment #4)
> ehsan sudden thought, wouldn't a backout of a cset that landed on trunk
> affecting .webidl be equally as painful for web compat and stability as a
> .webidl changing due to new functionality?

Yes, absolutely.

> As in, is this "really" the right thing here. (I'm just drawing attention to
> a case that I didn't see called out, if you already thought of it and this
> is still desired, great.)

Well, the assumption is that presumably a DOM peer has looked at the changes originally.  The fact that backouts often have to be done in a hurry while the trees are burning makes it impossible for us to impose more strict requirements on them.  :/