137583 - Contents of <URL: > not recognized as URL

Status: RESOLVED DUPLICATE of bug 5351

(Core :: Networking, defect)

Description

[Ralph Hartley]

RFC 1738 (appendix) recomends labeling URLs in documents, were they are not
otherwise obvious, with the prefix "URL:", and that they in addition should be
enclosed in "<" and ">".

Mozilla seems to work hard at finding unmarked URLs in text, but doesn't
recognize them, or correctly identify their boundaries, when they are labeld in
the recomended manner.

This is particularly a problem when this standard is used to prevent mangling by
line wrapping.

For example <URL:http://asg.web.
cmu.edu/rfc/rfc1738.html#page-22> 

(I broke the line on purpose, but often it is not under the users control.

I appologize if this is a dupe, but "URL" is not a good search term.

Comment 1

[benc]

+cc andreas for URL help.

There have been other bugs on URL's w/ leading and trailing whitespace.

Comment 2

[Andreas Otte]

ccing Ben, I think he wrote some stuff that recognizes URLs in Text.
(mozTXTToHTMLConv.cpp)

This has nothing to do with leading/trailing spaces.

Comment 3

[Ben Bucksch (:BenB)]

This is INVALID. The converter has a *special mode* just for this type of link
marking. It does recognize them correctly: <URL:foo:bar>.

The converter cannot recognize URLs in Mailnews that cross linebreaks because of
 a limitation of libmime. That's bug 5351.

For the record: RFC 1738 has been superceeded by RFC 2396, which does not
recommend this syntax anymore.

Comment 4

[Ben Bucksch (:BenB)]

> It does recognize them correctly: <URL:foo:bar>.

Ops, bad example. foo: is not a registered scheme in Mozilla and thus won't be
recognized. <URL:http://foo/bar> should work.

Comment 5

[Andreas Otte]

Ben, take a look at the test url: http://asg.web.cmu.edu/rfc/rfc1738.html#page-22

Seems to me, the code does not like whitespaces and linebreaks. I think you
should reconsider the INVALID, at least take a longer look at the examples.

Comment 6

[Ben Bucksch (:BenB)]

For a moment, I thought you were right, but then I noticed that it's
format=flowed. If the URL is broken by a space (seemingly) and not correctly
recognized, make sure that there's no linebreak in the message source.

Try sending yourself the following lines with the plaintext editor (SHIFT-click
on the composer, if you use the HTML composer by default):

foo <URL:ftp://info.cern.ch/pub/www/doc; type=d> foo
foo <URL:ftp://ds.in ternic.net/rfc> foo
foo <URL:http://ds.internic. net/instructions/overview.html#WARNING> foo

Then view the mail in Composer. All URLs get recognized correctly for me, with
whitespace removed.

Comment 7

[Andreas Otte]

Hmmmm ... so mailnews gets it right, except when the url is broken over multiple
lines because libmime gets one line after the other. This would make it a dupe
of bug 5351.

bugzilla also does not get it right with linebreaks and whitespaces. But that is
a bugzilla issue, see bug 137016 which has a patch.

Comment 8

[Ralph Hartley]

The behavior that made me submit this bug was caused by bug 5351
and the other examples were wrong, so I would call this a dup. (Reopening
so it can be resolved as dup)

I was confused by the examples in the rfc. They appear to be examples of use of the
recomended form which have been misinterpreted, but if you look at the source,
they have the incorrect interpretation built in. Maybe the document was damaged
when it was htmlified. Bugzilla did the same thing, reinforcing this illusion.

rfc2396 does not recomend "URL:" but still recomends "<" and ">".

The examples (without line breaks) do appear to be treated correctly in email.

Comment 9

[Ralph Hartley]

*** This bug has been marked as a duplicate of 5351 ***