575901 - DeCOMtaminate nsIStyleRuleProcessor method signatures

Status: RESOLVED FIXED

(Core :: CSS Parsing and Computation, defect)

Description

[Craig Topper]

Attachment: Patch

Converted an outparam on one method. The others I just changed NS_IMETHOD to virtual nsresult because the implementations can return something other than NS_OK. Though I don't think the caller of RulesMatching checks the result. So I'm willing to do further changes to remove the return type with some help.

Comment 1

[Boris Zbarsky [:bzbarsky]]

> Though I don't think the caller of RulesMatching checks the result.

Indeed.  I think we should nix those return types.  What help do you need?

Comment 2

[Craig Topper]

There's an OOM check in ElementTransitions::EnsureStyleRuleFor that causes RulesMatching to return NS_ERROR_OUT_OF_MEMORY. Can I just remove the OOM check from that method?

Comment 3

[Boris Zbarsky [:bzbarsky]]

Yes, operator new never returns null on trunk.

Comment 4

[Craig Topper]

Attachment: Remove nsresult return type from RulesMatching

Removed the OOM check so I could remove the nsresult return type.

Comment 5

[Craig Topper]

Attachment: Remove nsresult return type from RulesMatching updated to tip

Comment 6

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 455078 [details] [diff] [review]
Patch

r=bzbarsky

Comment 7

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 455891 [details] [diff] [review]
Remove nsresult return type from RulesMatching updated to tip

This is fine, but one followup that might be worth doing is to implement empty version of the various RulesMatching methods directly on nsIStyleRuleProcessor and remove all the empty subclass impls  Might make icache a little happier.

Comment 8

[Dão Gottwald [:dao]]

http://hg.mozilla.org/mozilla-central/rev/02e762e39d42