1959 - HTTP "Link" header prevents page from rendering!!!

Status: VERIFIED FIXED

(Core :: CSS Parsing and Computation, defect, P1)

Description

[Hixie (not reading bugmail)]

http://cvs-mirror.mozilla.org/webtools/bugzilla/enter_bug.cgi?product=NGLayoutThe relevant section of the HTML 4.0 spec is:  http://www.w3.org/TR/REC-html40/present/styles.html#h-14.6 and this refers to section 19.6.1.2 of the HTTP 1.1 RFC:  http://info.internet.isi.edu/in-notes/rfc/files/rfc2068.txt  But basically, the bug is that NGLayout fails to recognise the  HTTP header "link", which should be mapped into HTML LINK elements.  This is test 34 of the Import Test. If you have any questions about  the importtest, e-mail py8ieh=[e-mail]

Comment 1

[Hixie (not reading bugmail)]

Eek! My comments got mangled. Here they are again:


The relevant section of the HTML 4.0 spec is:
  http://www.w3.org/TR/REC-html40/present/styles.html#h-14.6
and this refers to section 19.6.1.2 of the HTTP 1.1 RFC:
  http://info.internet.isi.edu/in-notes/rfc/files/rfc2068.txt

But basically, the bug is that NGLayout fails to recognise the
HTTP header "link", which should be mapped into HTML LINK elements.

This is test 34 of the Import Test. If you have any questions about
the importtest, e-mail py8ieh=[e-mail]

Comment 2

[Hixie (not reading bugmail)]

Page moved, so correct the universal resource identifier.

Comment 3

[Hixie (not reading bugmail)]

Peter, I don't know what you did but now any page with an HTTP link header
just doesn't render.

That is, of all the pages on the ImportTest, any with a
   <META http-equiv="Link" ...>
...will simply stop downloading when they get to the first such construct,
and nothing will render. No error messages appear on the console, either.

This occurs with *all* "link" headers, not just stylesheet ones. It does
not occur with other HTTP headers such as Content-Base. It does not occur
on stress pages, i.e. the biggie altcascade.html (test 49) is passed by
NGLayout with flying colours (well, except that the UI cannot get to all
the stylesheets...).

This problem occurs on the following pages:
http://www.bath.ac.uk/%7Epy8ieh/internet/importtest/main/
http://www.bath.ac.uk/%7Epy8ieh/internet/importtest/extra/alternate.html
http://www.bath.ac.uk/%7Epy8ieh/internet/importtest/extra/httplink.html
http://www.bath.ac.uk/%7Epy8ieh/internet/importtest/extra/httpexoticmedia.html
http://www.bath.ac.uk/%7Epy8ieh/internet/eviltests/cascade/acidlinkcascade.html
http://www.bath.ac.uk/%7Epy8ieh/internet/importtest/extra/alternate.html
http://www.bath.ac.uk/%7Epy8ieh/a/internet/eviltests/link3.html
http://www.bath.ac.uk/%7Epy8ieh/a/internet/eviltests/link4.html

This bug is preventing me from giving you full credit for passing the
ImportTest - for your own sake, therefore, I suggest you fix it fast! :-)

Comment 4

[Peter Linss]

Fixed. There was an unitialized bool that would randomly block the parser (and
then never unblock it, since it didn't think it was blocked).

Comment 5

[Hixie (not reading bugmail)]

You now pass 100% of part I of the ImportTest.