46107 - Page renders browser unusable

Status: VERIFIED WONTFIX

(Core :: DOM: HTML Parser, defect, P2)

Description

[Braden]

Not sure what the component should be. Attempting to load this page appears to
render the application unusable. Doesn't actually bring down the app, but
effectively a crash. Page in question to follow as an attachment.

Comment 1

[Braden]

Attachment: Test case

Comment 2

[Braden]

Oh yeah... This is build 2000072108.

Comment 3

[Asa Dotzler [:asa]]

HTML element  the test case is this:

<select><option label="1">One<option label="2">Two<option label="3">Three</select>

Comment 4

[Hixie (not reading bugmail)]

There are two problems here.

1. The test case is confusing the hell out of our Strict DTD.
2. When a page cannot be displayed, Gecko doesn't draw anymore.

The second is almost certainly covered elsewhere (bug id anyone?) so lets 
focus on the first issue here.

-> Parser.

Nominating for nsbeta3 because of poor user experience. This is not 
a 'correctness' issue -- the HTML is invalid. It's not a crash, the browser 
stands solid but gecko stops drawing. It's not a leak, it's not a ui issue,
it's not a polish issue (it's rather more important than a polish issue!)...

Comment 5

[Braden]

Yup... I stumbled onto this one by accident. :-)

I don't know what the verdict is as far as the browser's response to invalid
documents that claim to be 4.0 Strict--IMO, an error message would be
sufficient. In other words, I don't know that 1 is a problem. As far as 2 goes,
the current behavior ("freak out") is obviously unacceptable. :-)

If 2 is documented elsewhere and there is agreement that 1 isn't a problem, this
can probably be marked as a duplicate.

Comment 6

[Hixie (not reading bugmail)]

The first is a problem because even with blank documents, the parser should be
creating a content model. With this page, it freaks out completely, which is
wrong wrong wrong. (Of course, there is also the additional problem of what to
do when we *do* freak out, but that is a separate issue.)

Comment 7

[harishd]

Assigning to myself since rickg is on sabbatical.

Comment 8

[Nisheeth Ranjan]

This should be an easy fix according to Harish.  We will make it so that the 
parser always opens up a body so that a blank page gets displayed in such cases.

Comment 9

[harishd]

Setting priority to P2.

Comment 10

[rickg]

Marking wont fix -- we're removing strictDTD support in 6.0.

Comment 11

[Jan Carpenter]

verified