[Bug 21945] Space should be allowed between -- and > in comment end

[bugzill...@webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=21945





--- Comment #26 from Julien Chaffraix <jchaf...@webkit.org> 2010-05-12 08:38:08 PST ---
(In reply to comment #25)
> Does HTML5 need to be corrected?

It does not. You can just discard my previous comment about those states not being reached and the dashes being consumed. After sleeping on that, we need to implement the 2 states to close such comments: <!--> or <!--->. I will add some testing for those and post the updated patch soon.

--
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.

[bugzill...@webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=21945


Julien Chaffraix <jchaf...@webkit.org> changed:

What |Removed |Added
----------------------------------------------------------------------------
Attachment #55857| |review?, commit-queue-
Flag| |




--- Comment #27 from Julien Chaffraix <jchaf...@webkit.org> 2010-05-12 09:24:54 PST ---
Created an attachment (id=55857)
--> (https://bugs.webkit.org/attachment.cgi?id=55857)
Updated HTML5 parsing - added the 2 missing states, cleaned some more old code and this time with the test cases

[bugzill...@webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=21945


Alexey Proskuryakov <a...@webkit.org> changed:

What |Removed |Added
----------------------------------------------------------------------------
Summary|Space should be allowed |Implement HTML5 comment
|between -- and > in comment |parsing
|end |




--- Comment #28 from Alexey Proskuryakov <a...@webkit.org> 2010-05-12 09:26:29 PST ---
Renaming to match the scope of work in this fix.

I'd appreciate it if you could post an overview of where we'll stand compatibility-wise. My understanding is that we'll match HTML5, but what the new differences with IE and with Firefox will be?

[bugzill...@webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=21945





--- Comment #29 from Julien Chaffraix <jchaf...@webkit.org> 2010-05-14 07:39:23 PST ---

> I'd appreciate it if you could post an overview of where we'll stand compatibility-wise. My understanding is that we'll match HTML5, but what the new differences with IE and with Firefox will be?

Your understanding is right: we match HTML5 and will match the next release of Gecko on most cases.

- Compared to IE:
* we accept white space in the comment end (which was what this bug is about)

- Compared to Grand Paradiso (FF nightly as it does not make sense to compare to FF without the HTML5 parsing model):
* FF is more lenient as to when closing the comment (--ouch> does close a comment)
* FF has a bug where the comment end delimiter is not properly consumed (filed https://bugzilla.mozilla.org/show_bug.cgi?id=565896 about that)
* looking at the code, they are a bit more lenient in their parsing. One example is that they are accepting '\r' in the comment end delimiter.

- Common:
* we emit comments without an end delimiter (as tested in the new tests in the patch) whereas IE and FF just don't recognize them as comments. Not sure if we should amend HTML5 or file a Mozilla bug about that.

[bugzill...@webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=21945





--- Comment #30 from Julien Chaffraix <jchaf...@webkit.org> 2010-05-14 09:07:18 PST ---
I have to amend FF results as I checked the wrong nightly (3.0.19pre vs 3.7):

> - Compared to Grand Paradiso (FF nightly as it does not make sense to compare to FF without the HTML5 parsing model):
> * FF is more lenient as to when closing the comment (--ouch> does close a comment)
> * FF has a bug where the comment end delimiter is not properly consumed (filed https://bugzilla.mozilla.org/show_bug.cgi?id=565896 about that)

Those have been solved.

> * looking at the code, they are a bit more lenient in their parsing. One example is that they are accepting '\r' in the comment end delimiter.

This still holds as it was code inspection.

New difference:

* FF does consider parse comments in <script>, <style> and <title> as if it was HTML.

This is a difference with our parsing model.

> - Common:
> * we emit comments without an end delimiter (as tested in the new tests in the patch) whereas IE and FF just don't recognize them as comments. Not sure if we should amend HTML5 or file a Mozilla bug about that.

We match FF on this one so they emit comment if they reach EOF.