127071 - Remove AREA and MAP specific code from parser

Status: RESOLVED FIXED

(Core :: DOM: HTML Parser, defect, P5)

Description

[Fabian Guisset]

From Bugzilla Helper:
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.0; en-US; rv:0.9.8) Gecko/20020204
BuildID:    20020221

We used to do some special magic to AREA's and MAP's in the parser when building
the content model, but that time is over. Gotta remove
HTMLContentSink::ProcessMAPTag, HTMLContentSink::ProcessAREATag,
nsIHTMLContentSink::OpenMap, nsIHTMLContentSink::CloseMap. The patch still needs
some cleanup, but I'll attach it soon enough.
Targetting for 1.0 even though I doubt the drivers will like it.

Reproducible: Always
Steps to Reproduce:
1. Open nsHTMLContentSink.cpp
2. Wonder why there is AREA and MAP specific code
3.

Actual Results:  Bleh

Expected Results:  Blah

nsHTMLContentSink.cpp needs a spanking. Thus spoke Zaratustra.

Comment 1

[Fabian Guisset]

Attachment: Off with their head.

This should do it. Tested and image maps still work fine.
HOWEVER I have no idea how to test the other sinks, nsLoggingSink, nsRobotSink,
and nsHTMLNullSink. I'd really like an approval from harishd before I even
consider checking this in.

Comment 2

[harishd]

nsLoggingSink is used for regression testing. I've never used
nsHTMLNullSink/nsRobotSink. Anyway, before I review your patch let me make sure
that it doesn't break regression testing.

Comment 3

[Fabian Guisset]

Cleanup, non-critical -> Post 1.0 job. Waiting for harish's regression tests
results, but we're not in a hurry :-)

Comment 4

[Fabian Guisset]

Future, unless someone updates the patch (I have no idea if it's still needed)

Comment 5

[John Keiser (jkeiser)]

Comment on attachment 70974 [details] [diff] [review]
Off with their head.

Harish?  I really like the number of minuses in this patch and we still seem to
have those methods ...

Comment 6

[harishd]

Fabian, have you tested nested MAPs? I'm asking this question because the NavDTD
uses a counter ( mOpenMapCount ) to make sure that nested MAPs work correctly
and I'm a bit concerned that your patch might break that.

Comment 7

[Fabian Guisset]

That's a good point I've not tested nested maps I didn't even know it was
possible :-)
Anyway I am currently unable to do any work on Mozilla (no computer) so feel
free to drive the patch into the tree if you really want it :-)

Comment 8

[harishd]

Comment on attachment 70974 [details] [diff] [review]
Off with their head.

We need to make sure that nested MAPs don't break.

Comment 9

[Boris Zbarsky [:bzbarsky]]

peterv, care to pick this up?

Comment 10

[:Ms2ger (he/him; ⌚ UTC+1/+2)]

Fixed by

Bug 312592 - Reduce content creation redundancy in nsHTMLContentSink
Bug 311785 - Simplify nsIHTMLContentSink

in late 2005.