317254 - [FIX]Redirect (meta refresh) doesn't escape URL, seems to do a lossy UTF-16 to ASCII conversion instead

Status: RESOLVED FIXED

(Core :: DOM: Core & HTML, defect)

Description

[Jesse Ruderman]

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.8) Gecko/20051111 Firefox/1.5

Steps to reproduce:
1. Load http://www.linuxshare.ru/tmp/ffox/index.cgi?item=%F1%CE%D5%D3
2. Wait for the page to refresh.

Result: Firefox 1.5RC3 loads http://www.linuxshare.ru/tmp/ffox/index.cgi?item=/=CA.  That's obviously wrong: it looks like Firefox started to send the URL as UTF-16 without escaping, then sent the URL through a lossy UTF-16-to-ASCII converter.

Expected: Load http://www.linuxshare.ru/tmp/ffox/index.cgi?item=%F1%CE%D5%D3 again (I think).

In contrast, Firefox 1.0.7 loads http://www.linuxshare.ru/tmp/ffox/index.cgi?item=%D0%AF%D0%BD%D1%83%D1%81, which is correct except that it's encoded as UTF-8 instead of what the site expects (KOI8-R, I think).  Despite that, I think there's some way this breaks sites that worked correctly in Firefox 1.0.7 based on the way Igor N. Avtaev described the bug.

Originally reported by Igor N. Avtaev:
http://www.squarefree.com/burningedge/2005/11/11/2005-11-11-branch-respin/#comment-3037
http://www.squarefree.com/burningedge/2005/11/11/2005-11-11-branch-respin/#comment-3057

Comment 1

[Simon Montagu :smontagu]

http://bonsai.mozilla.org/cvsblame.cgi?file=mozilla/content/base/src/nsDocument.cpp&rev=3.604&mark=1386-1387#1375

Comment 2

[Masayuki Nakano [:masayuki] (he/him)(JST, +0900)]

Note that this testcase is INVALID HTML document. On HTML document, we cannot use non-ASCII character for URI values. The HTML author MUST use escaped URI in URI values.

But of course, we need to fix this quirks if we can do it.

Comment 3

[Darin Fisher]

-> default owner

Comment 4

[Mike Kaply [:mkaply]]

Is there any interest in taking a fix for this regression?

Is this in the right component?

Comment 5

[Boris Zbarsky [:bzbarsky]]

Yes and "probably not given the link in comment 1".

We used to just convert the URI to UTF8.  Maybe we should go back to doing that.  

Comment 6

[Boris Zbarsky [:bzbarsky]]

Attachment: Like so

Comment 7

[Boris Zbarsky [:bzbarsky]]

This just restores the old behavior.  We _could_ try to encode using the page charset, I suppose, but I'm not sure we want to.

Comment 8

[Johnny Stenback (:jst)]

Comment on attachment 281327 [details] [diff] [review]
Like so

r+sr=jst

Comment 9

[Boris Zbarsky [:bzbarsky]]

Comment on attachment 281327 [details] [diff] [review]
Like so

Requesting approval.  This should be a very low-risk change to restore behavior we used to have back in the 1.7 timeframe.

Comment 10

[Boris Zbarsky [:bzbarsky]]

Checked in.  We still need a test, though.  I could use some help writing one, to be honest.

Comment 11

[Jonas Sicking (:sicking) No longer reading bugmail consistently]

Marking as fixed, we have the in-testsuite flag to track the lack of test. Please reopen if there are other issues remaining.

Comment 12

[Boris Zbarsky [:bzbarsky]]

This should have been marked fixed, yeah.  Not sure why that failed.