848265 - No styling in ChatZilla user list

Status: VERIFIED FIXED

(Other Applications :: ChatZilla, defect)

Description

[Tony Mechelynck [:tonymec]]

Last known good nightly:
Mozilla/5.0 (X11; Linux x86_64; rv:22.0) Gecko/20100101 Firefox/22.0 SeaMonkey/2.19a1 ID:20130304003003 c-c:3044a67220c7 m-c:f99a075a5bce

Both yesterday's and today's nightlies are "bad".
Same ChatZilla version, 0.9.90
Same ChatZilla motif, http://chatzilla.hacksrus.com/motifs/output-light-faces.css (the "output-light" motif without "faces" has the same problem)

Symptoms:
No styling in userlist, i.e. no italics, no grey text and no @#+ etc. symbols in front of privileged users. The list *is* sorted by user status.

Bug is found in both SeaMonkey and Firefox. I'm unable to test other platforms.

Message in Error Console (repeated many times):
Security Error: Content at http://chatzilla.hacksrus.com/motifs/output-base.css may not load or link to chrome://chatzilla/content/output-base.css.

There are other messages, warnings and errors, even for ChatZilla; this one seems most relevant.

I'm filing this bug in ChatZilla rather than in Core because this message seems to indicate that ChatZilla has been hit by a new irreversible turn of the screw in Core security. I'll try to download the motif and see if it changes anything. If it does, then I suppose that (unless some other way is found to circumvent the chrome/content privilege barrier) the standard "light" and "dark" motifs will have to be distributed as part of the ChatZilla chrome.

Comment 1

[James Ross]

I'd guess it's just bug 407956 landing, but it could be a security thing.

Comment 2

[Tony Mechelynck [:tonymec]]

After doing the following, the bug is still present in today's nightly:
1. download output-light-faces.css, output-light.css and output-base.css from http://chatzilla.hacksrus.com/motifs/ to a local directory
2. replace all relative URLs to images in the local output-light-faces by absolute URLs
3. change the "current motif" in ChatZilla global appearance preferences to a file:/// URL pointing to the local output-light-faces

(In reply to James Ross from comment #1)
> I'd guess it's just bug 407956 landing, but it could be a security thing.
That bug landed (by several changesets, but I suppose the relevant one is mozilla-central changeset bf88a316cf18) on April 4, and it has “addon-compat, dev-doc-needed” keywords. Yes, it sounds like, at the minimum, something to be looked into.

Comment 3

[James Ross]

If you select the light or dark motifs from the View > Color Scheme menu, you get chrome: URIs FWIW. That should completely rule out security.

Comment 4

[Tony Mechelynck [:tonymec]]

(In reply to James Ross from comment #3)
> If you select the light or dark motifs from the View > Color Scheme menu,
> you get chrome: URIs FWIW.

ah, yes indeed.

> That should completely rule out security.

After "View → Color Scheme → Light Motif" the bug is still there.

Comment 5

[Szabolcs Hubai (:xabolcs)]

Attachment: Address nsITreeView changes - v1

Tested with "unpack-edit-pack-restart" method.

In ul_getcellprops I tried to avoid code duplicating,
therefore I created the local variable resultProps array.

Comment 6

[James Ross]

Comment on attachment 725942 [details] [diff] [review]
Address nsITreeView changes - v1

Review of attachment 725942 [details] [diff] [review]:
-----------------------------------------------------------------

Looks good, thanks!

Comment 7

[James Ross]

http://hg.mozilla.org/chatzilla/file/01bcf4d6b7a2

Comment 8

[Tony Mechelynck [:tonymec]]

Mozilla/5.0 (X11; Linux x86_64; rv:22.0) Gecko/20100101 Firefox/22.0 SeaMonkey/2.19a1 ID:20130331171202 c-c:b8bc5996f9ba m-c:0b7c27024048

After doing the following:
1. Download this hourly
2. Untar it in /usr/local
3. Start it.
4. Browse to file:///usr/local/seamonkey/distribution/extensions/
5. Click on {59c81df5-4b7a-477b-912d-4e0fdf64e5f2}.xpi and accept the install
6. Restart (with Browser, Mail and ChatZilla)

- I see @, % and + icons next to nicknames again
- After a second or two (the time required to query the server I suppose) nicknames of /away users are greyed-out and in italics.

This is whatever ChatZilla 0.9.90 source was at the head of the default branch in the ChatZilla repository at 2013-03-31 17:12:02 PDT, which is later than comment #7.

Thanks xabolcs for the coding and Silver for reviewing and pushing!

Comment 9

[Mark Capella [:capella]]

So is this finished? How does a user (me for example :) ) get the new patch into his currently installed FF / WIN / v0.9.90 ?

Comment 10

[Tony Mechelynck [:tonymec]]

In reply to comment #9: Yes, it is finished. The source has been patched and all future builds, in particular the ChatZilla versions shipped with all current and future builds of SeaMonkey 2.20a1, will include the fix.

For Firefox, I don't know the "official" way to install the fixed version of ChatZilla (other than waiting for the next version of cZ to appear at AMO, of course). One possible way (but not the most elegant one, I'm sure) is to install SeaMonkey *2.20a1* (see below) on the same machine, then use Firefox to browse to file:///(something)/distribution/extensions and click {59c81df5-4b7a-477b-912d-4e0fdf64e5f2}.xpi to install it. The value of (something) is the platform-dependent SeaMonkey install directory; for instance typically usr/local/seamonkey on Linux. On Windows it might (no guarantee here) be something like c:/Program%20Files%20%28x86%29/Mozilla/SeaMonkey (note the forward slashes and the % hex escaping).

SeaMonkey 2.20a1 is the bleeding-edge development version of SeaMonkey (homologous with Firefox 23.0a1 Nightly); it can be downloaded (at your own risk) from http://ftp.mozilla.org/pub/mozilla.org/seamonkey/nightly/latest-comm-central-trunk/

Comment 11

[Mardeg]

(In reply to Mark Capella [:capella] from comment #9)
> So is this finished? How does a user (me for example :) ) get the new patch
> into his currently installed FF / WIN / v0.9.90 ?
Just install it from 
http://james-ross.co.uk/mozilla/chatzilla/nightly/builds/0.9.90/2013040112/chatzilla-0.9.90-2013040112.xpi

Comment 13

[Gene Scharmann]

(In reply to Mardeg from comment #11)
.....> Just install it from 
> http://james-ross.co.uk/mozilla/chatzilla/nightly/builds/0.9.90/2013040112/
> chatzilla-0.9.90-2013040112.xpi

Thanks, this version of Chatzilla fixed the problem which began when FF v22 was installed.

Comment 14

[Robert Utasi [:hunboy]]

Silver, can you reopen this bug? It still occurs, when we change the motif manually between light and dark motif. The background color doesn't change for userlist (and channel-tree plugin. note: I've tested in a clean install without plugin and also buggy, so not the channel-tree plugin does troubles)

It's already ok when we start the chatzilla with dark motif - everything is good. Occurs only when we change the motif.

Szabolcs, do you have an idea how to fix it?

Comment 15

[James Ross]

Changing motifs not taking effect is a different Firefox bug, unrelated to this one, AFAIK.

Comment 16

[Szabolcs Hubai (:xabolcs)]

(In reply to Robert Utasi [:hunboy] from comment #14)
> <snip>
>
> It's already ok when we start the chatzilla with dark motif - everything is
> good. Occurs only when we change the motif.
> 
> Szabolcs, do you have an idea how to fix it?

I can reproduce your issue with ChatZilla 0.9.86.1 [Iceweasel 17.0.7/20130626024843].
It's clearly not a regression caused by this bug. :)

Please file a new bug for it, and let's see it's a bug or an expected behavior!

Comment 17

[Robert Utasi [:hunboy]]

Thanks, Szabolcs. Me neither think if this patch causes this issue, just thought something similar, while this is the XUL Tree backgound, and this is stylesheet problem also and whilst you are in this bug by mind, did it worth 1 question maybe you had a fast idea. When I will have enough time I'll do regression range to figure it out where is it from.

Comment 18

[Szabolcs Hubai (:xabolcs)]

(In reply to Robert Utasi [:hunboy] from comment #17)
> Thanks, Szabolcs. Me neither think if this patch causes this issue, just
> thought something similar, while this is the XUL Tree backgound, and this is
> stylesheet problem also and whilst you are in this bug by mind, did it worth
> 1 question maybe you had a fast idea. When I will have enough time I'll do
> regression range to figure it out where is it from.

Sorry, I have no idea, how it can be fixed, or it's possible to fix at all.
But please find the regression range if any, and CC me on that bug! :)