428886 - enhancement: don't ignore newlines in HTML tooltips

Status: RESOLVED DUPLICATE of bug 358452

(Firefox :: General, enhancement, P4)

Description

[Amnon Aaronsohn]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.12) Gecko/20060601 Firefox/2.0.0.12 (Ubuntu-edgy)
Build Identifier: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9b3pre) Gecko/2008011304 Minefield/3.0b3pre

According to the HTML standard, in the TITLE attribute, CRs should be converted to spaces while LFs should be discarded. This has been implemented in trunk (bug 218223). However, I propose to ignore the standard this time because:

(1) this causes documents that are identical except for newline conventions (LF or CR/LF) to appear differently -- doesn't make sense

(2) it causes words that are separated by whitespace in the source to be not separated in the rendered document -- also doesn't make sense

So I'm asking that LF characters be converted to spaces along with the other whitespace chars.

Example:
<a title="simple
test">test</a>
shows as "simpletest" with UNIX newlines and "simple test" with DOS newlines.

Reproducible: Always

Comment 1

[Amnon Aaronsohn]

Attachment: convert LF chars to space instead of ignoring them

Comment 2

[Mats Palmgren (inactive)]

The HTML5 draft is very clear that LF "represents a line break":
http://www.w3.org/html/wg/html5/#the-title

> (LF or CR/LF) to appear differently -- doesn't make sense

FWIW, I agree, but I think we should make a line break for any combination,
for compatibility with other popular browsers (IE and Safari).

Comment 3

[Lucas Malor (mail: c6kfnkn2uc AT snkmail DOT c0m)]

We can't ask to W3C to change this behaviour? It obviously doesn't make any sense.

Comment 4

[jason.ratcliffe]

Just to say, I've found this to be an issue as well.

Comment 5

[:Gavin Sharp [email: [e-mail]]]

Comment on attachment 315465 [details] [diff] [review]
convert LF chars to space instead of ignoring them

Apologies for the long delay in reviewing here...

Making this change seems reasonable, but I'm reluctant to do it without comparing our behavior to the other mainstream browsers (Safari, Opera, IE, etc.), ideally on as wide a range of testcases as possible. Bug 218223 includes some testcases that might be useful.