501375 - nsHTMLTableAccessible::HasDescendant is much slower than it should be

Status: RESOLVED FIXED

(Core :: Disability Access APIs, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Obvious issues:

1)  It forces a walk over all of the table's kids by getting .length on the
    list.  For cases when the table actually _has_ a kid of the sort this
    algorithm is looking for, this is wholly unnecessary.
2)  It has a bizarre check for whether the first item of the list is the table
    itself.  This can't ever happen.
3)  It does some weird string operations instead of just having the callers pass
    in an NS_LITERAL_STRING().

Issue #1 is by far the most important, of course.

Comment 1

[alexander :surkov (:asurkov)]

Attachment: patch

not tested, it would be nice to get mochitests from bug 495388.

Comment 2

[Boris Zbarsky [:bzbarsky]]

That patch changed behavior such that if aAllowEmpty is false and the first kid is empty the function will return false where it used to return true....

Comment 3

[alexander :surkov (:asurkov)]

(In reply to comment #2)
> That patch changed behavior such that if aAllowEmpty is false and the first kid
> is empty the function will return false where it used to return true....

Ah, I see, thank you for the catch.

Comment 4

[alexander :surkov (:asurkov)]

(In reply to comment #2)
> That patch changed behavior such that if aAllowEmpty is false and the first kid
> is empty the function will return false where it used to return true....

Sorry I was quick to say I see. Do you mean the case when length > 1?

Comment 5

[Boris Zbarsky [:bzbarsky]]

Er, yes.  Sorry, forgot to mention that.

Comment 6

[alexander :surkov (:asurkov)]

Attachment: patch2

Comment 7

[Marco Zehe (:MarcoZ)]

Comment on attachment 386795 [details] [diff] [review]
patch2

r=me, thanks!

Comment 8

[alexander :surkov (:asurkov)]

checked in on 1.9.2 - http://hg.mozilla.org/mozilla-central/rev/7ed7057a36c7