178461 - nsIXSLTProcessor::reset()

Status: VERIFIED FIXED

(Core :: XSLT, enhancement)

Description

[Axel Hecht]

XSLTProcessor should have a reset method to prune params and stylesheets.
Applications like buster shouldn't stress test the js garbage collection for
something that can be done that easy.

Comment 1

[Axel Hecht]

Attachment: implement reset()

Comment 2

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Comment on attachment 105193 [details] [diff] [review]
implement reset()

personally i think that we should have a ::ResetStylesheets rather then a Reset
function. That way there is no duplication of functionality in two separate
functions in the interface. And we'd support dropping just the stylesheet in
case someone wants to use the same parameters for different stylesheets (though
that is probably an edgecase, on the otoh using the processor the way that
buster is is an edgecase as well).

However if you guys decide that we should have a ::Reset then r=me

Comment 3

[Axel Hecht]

I see alot more scenarios that include reseting both stylesheets and params than
keeping just the params.
The real issue I had was "make that processor a virgin again" (which works much
better for code than for girls). Keep the common things simple, and make uncommon
things possible. So if somebody wants the params and a different stylsheet, (s)he
can just set the params again.

Comment 4

[Peter Van der Beken [:peterv]]

Comment on attachment 105193 [details] [diff] [review]
implement reset()

Safe sex is good! sr=peterv.

Comment 5

[Axel Hecht]

checked into the trunk, is this sexy enough to go into 1.2?

Comment 6

[Axel Hecht]

Comment on attachment 105193 [details] [diff] [review]
implement reset()

requested approval for 1.2

Having .reset() on that milestone will ease the move to the new interface, as
we have only one version of it out there, as agreed on with peterv on irc

Comment 7

[Asa Dotzler [:asa]]

Comment on attachment 105193 [details] [diff] [review]
implement reset()

we're down to taking only crash fixes, major regression fixes or highly visible
bugs. This seems more featurish so I'm saying no to 1.2.

Comment 8

[Axel Hecht]

fixed in 1.3a, not fixed in 1.2 per drivers

Comment 9

[Axel Hecht]

mass verifying