397021 - Subject line <foo@bar> in the forward inline body not shown in html composition mode

Status: RESOLVED FIXED

(MailNews Core :: MIME, defect)

Description

[Jonathan Kamens]

User-Agent:       Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8.1.6) Gecko/20070911 Fedora/2.0.0.6-8.fc8 Firefox/2.0.0.6
Build Identifier: version 2.0.0.6 (20070911) from Fedora Rawhide

If I create a message containing subject"<foo@bar>" in the subject, send it to myself, and then try to forward it, the "<foo@bar>" is stripped from the subject.  It's just wrong for thunderbird to be mucking with subjects like that.


Reproducible: Always

Comment 1

[Magnus Melin [:mkmelin]]

I assume you mean in the body where the --- Original Message --- stuff is. Confirming on linux/trunk.

Comment 2

[Magnus Melin [:mkmelin]]

Or rather, it's just not shown (in html mode) as the < > are not encoded as &lt; and &gt; -- or so I assume as the plain text composition works fine.

Comment 3

[Jonathan Kamens]

Attachment: proposed fix

Thanks for fixing this bug so quickly. :-)
Patch attached.

Comment 4

[Ludovic Hirlimann [:Usul]]

(In reply to comment #3)
> Created attachment 460147 [details] [diff] [review]
> proposed fix
> 
> Thanks for fixing this bug so quickly. :-)
> Patch attached.

You need to request reviews if you want you code to get in see https://developer.mozilla.org/en/Mailnews_and_Mail_code_review_requirements.

Comment 5

[Andrew Sutherland [:asuth] (he/him)]

Comment on attachment 460147 [details] [diff] [review]
proposed fix

Ludo, I'm not a super-reviewer although I've heard talk of trying to make that happen.

But I can chime in here as a reviewer and say that this definitely needs at least a unit test confirming the change in behaviour.  If there are not already unit tests covering the cases where the removed/modified logic was supposed to be doing something helpful, test cases should probably be added for that too.  (Please call out the existing tests in a buzilla comment if they already exist.)

Sorta helpful testing link:
https://developer.mozilla.org/en/MailNews_xpcshell-tests

bienvenu is likely the right person to ask for review from once tests are created or if help is required in creating tests.  I don't think super-review is required for such a patch, but if it is, bienvenu can give it.

Comment 6

[Ludovic Hirlimann [:Usul]]

(In reply to comment #3)
> Created attachment 460147 [details] [diff] [review]
> proposed fix
> 
> Thanks for fixing this bug so quickly. :-)
> Patch attached.

Jonathan , you'll need to add some unit test (I forgot to look for these). If you need help with these please ask either here or on our mailing-lists or in #maildev on irc.mozilla.org.

Comment 7

[Jonathan Kamens]

The learning curve for figuring out how to debug and fix Thunderbird issues was steep enough. Unfortunately, I do not have the additional time that would be necessary to get over the undoubtedly steep learning curve that would also be necessary to write a unit test for something like this.
One would hope that if a user of the software gave the time and energy to submit a fix, the developers of the software would meet the user halfway and write the unit test. Especially given that this bug has sat unfixed for almost three years. In short, can I please get a little help here?
I've owned free software projects for much longer than Mozilla has existed, and I've never asked a user who contributed a fix to write a unit test for it.

Comment 8

[Jonathan Kamens]

Is there no one willing to meet me halfway here and write a unit test for this bug? I'd really appreciate the help.

Comment 9

[Jonathan Kamens]

Since posting comment #7 and comment #8, I've learned how to build and patch TB and FF and how to work with the unit tests for both, so I'll write my own unit test for this bug when I get a chance. That may be a while, though -- too many all-nighters fixing other TB bugs have left me with a work deficit I need to address -- so if anyone else stumbles across this bug and wants to pitch in by writing the unit test, I wouldn't say no.

Comment 10

[Magnus Melin [:mkmelin]]

Attachment: mozmill test

Here's a mozmill test for it, you'd run it using 
make SOLO_TEST=composition/test-forwarded-content.js mozmill-one

Comment 11

[Jonathan Kamens]

Attachment: Consolidated patch

Thank you Magnus! Here is a single consolidated patch that incorporates my fix and Magnus's mozmill test. Review?

Comment 12

[David :Bienvenu]

Comment on attachment 609755 [details] [diff] [review]
Consolidated patch

thx for the patch! The mime parts of the patch seem to have bit-rotted - can you attach a de-bitrotted patch and re-request review? thx!

Comment 13

[Jonathan Kamens]

Attachment: Consolidated patch without repetition

Apparently "hg export -o" appends instead of overwriting. D'oh! Updated patch attached.

Comment 14

[David :Bienvenu]

Comment on attachment 610139 [details] [diff] [review]
Consolidated patch without repetition

thx for the patch, looks reasonable - can you use nsnull instead of NULL?

Comment 15

[Jonathan Kamens]

Attachment: Consolidated patch that uses nsnull instead of NULL

Same as attachment 610139 [details] [diff] [review] which is already review+. The only change is to replace NULL with nsnull as requested by the reviewer. I've confirmed that it builds with the nsnull change, so I'm carrying forward the review+.

Comment 16

[Ryan VanderMeulen [:RyanVM]]

http://hg.mozilla.org/comm-central/rev/30c2149ba887

Comment 17

[[e-mail]]

Bah, you're supposed to use MsgEscapeHTML, as per line 1435... patch coming up.

Comment 18

[[e-mail]]

Attachment: Use MsgEscapeHTML

Comment 19

[Jonathan Kamens]

Sorry, didn't know.

Why do we have two functions that do exactly the same thing?

Comment 20

[David :Bienvenu]

Comment on attachment 610985 [details] [diff] [review]
Use MsgEscapeHTML

Neil's talking about supporting frozen linkage, which is a different way of building where only some interfaces are supported.

Comment 21

[[e-mail]]

Comment on attachment 610985 [details] [diff] [review]
Use MsgEscapeHTML

Pushed changeset b1b2d61638f5 to comm-central.