501978 - reply list broken for "List-Post: <mailto:foo@bar> (Comment)"

Status: RESOLVED FIXED

(Thunderbird :: Mail Window Front End, defect)

Description

[Magnus Melin [:mkmelin]]

Attachment: proposed fix

Regressed with bug 496439.

If you have a header
List-Post: <mailto:foo@bar> (Post new message to the foo bar mailing list)

... the reply to list won't be enabled.

Comment 1

[Phil Ringnalda (:philor)]

Isn't that just going to let you use reply to list to open a compose window without an address to send to? The point of the regex was that because compose uses an incorrect heuristic, only adding the list-post to the To field if (StringBeginsWith(listPost, NS_LITERAL_STRING("<mailto:")) && listPost.CharAt(listPost.Length() - 1) == '>'), we shouldn't be saying that you can reply-to-list when we're going to screw it up.

Comment 2

[Magnus Melin [:mkmelin]]

Ah yes :(

Comment 3

[Magnus Melin [:mkmelin]]

Attachment: proposed fix, v2

Well this makes the backend work too. 
I'm not sure if it qualifies as the parser bug 496914 wants...

Comment 4

[Phil Ringnalda (:philor)]

It might be good enough to go on with (I'd really love it if one of these bugs had an mbox attachment with a whole bunch of possible list-post headers, if not an actual test), but it's not the parser in question, since that would

* strip leading whitespace and comments, then if the first character isn't "<" discard the whole field

* strip comments (to not get fooled by one containing a comma), then

* split(",")

* if any chunk does not contain a <...>, discard it and any following it

* if any chunk contains anything but whitespace following the ">", discard it and any following chunks

Most of the things where you would find a list-post address you shouldn't would be just uninteresting spec-compliance failures like not knowing how to comment - the only "real" one I can think of offhand would be "NO (used to be <mailto:[e-mail]> but we now treat that as a honeypot for our blackhole)".

Comment 5

[Magnus Melin [:mkmelin]]

Attachment: proposed fix, v3

bwinton pointed out the substring was wrong

Comment 6

[Blake Winton (:bwinton) (:☕️)]

Attachment: An mbox for philor.

(In reply to comment #4)
> (I'd really love it if one of these bugs had an mbox attachment with a whole
> bunch of possible list-post headers, if not an actual test),

Okay, here's your mbox attachment.  It's got a couple of List-Post headers, ("NO", "[e-mail]", and "<mailto:[e-mail]>", which should cover this bug), and some weird To: headers, as well, if you wanted to use it in a different bug.

As a side note, all of these messages come from the wild.  Yes, even the "To: A Group:Chris Jones <[e-mail]>,[e-mail],John <[e-mail]>;"  :)

Later,
Blake.

Comment 7

[Phil Ringnalda (:philor)]

Attachment: A bit more testcasey

This isn't quite complete, but covers a few more things (and doesn't make me think about what's expected, always a plus in a testcase), including several we'll still fail on, none of which excite me quite enough to say we have to fix them here.

Comment 8

[Phil Ringnalda (:philor)]

Comment on attachment 386580 [details] [diff] [review]
proposed fix, v3

r=me, and somebody pretty please write a real parser before we touch this again!

Comment 9

[Magnus Melin [:mkmelin]]

changeset:   3336:bcf6c88c8d73
http://hg.mozilla.org/comm-central/rev/bcf6c88c8d73

->FIXED