312474 - IMAP/RSS message headers not shown if there is no POP-mail account set up

Status: RESOLVED FIXED

(Thunderbird :: Mail Window Front End, defect)

Description

[Matthew Daniel]

User-Agent:       Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9a1) Gecko/20051014 Firefox/1.6a1
Build Identifier: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9a1) Gecko/20051014 Firefox/1.6a1

With the feed set to "Show article summary instead of web page" it used to show
the subject and (most importantly) the URL of the article being summarized. Now
with 1.5b2 and with HEAD this is no longer true. It shows no headers at all.

This may be releated to Bug 273687 but it was opened a LONG time ago.

Reproducible: Always

Steps to Reproduce:
1. Download TB 1.5b2
2. rm -rf ~/.thunderbird
3. Subscribe to http://download.freshmeat.net/backend/fm-releases-unix.xml
4. Ensure that "Show article summary" is checked
5. Open any article
6. Observe that the body is presented, but the only way to find the source
artitle is View Source (not even "View, Headers, All" helps).

Comment 1

[Magnus Melin [:mkmelin]]

"Website" show just fine over here.
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b5) Gecko/20051014
Thunderbird/1.4.1 ID:2005101404

Comment 2

[Matthew Daniel]

(In reply to comment #1)
> "Website" show just fine over here.
> Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b5) Gecko/20051014
> Thunderbird/1.4.1 ID:2005101404

Sure, that's why I said it _used_ to show the website, back in the "old days."
With the latest release generally available from mozilla.org, it does not.

I admit I should have picked 1.5a2 as the "Version" field of the Bug, but it
applies to "HEAD", too. As I clearly stated in my initial report.

I'll switch the Version to help eliminate further confusion. Thanks for checking
into this.

Comment 3

[Magnus Melin [:mkmelin]]

I was testing with the latest 1.5 branch nigtly (=1.5b2+some bug fixes), and
"website" does show for me. 

Comment 4

[Matthew Daniel]

Attachment: A small screen-cap to clarify some terms

I hoped maybe we were just talking about two separate things, instead of the
far worse thought that our builds are actually different. I tried it at home on
x86 (my work, and the build that originated the bug, is x86_64) and it does the
same thing with a build yesterday afternoon.

Please tell me how I can help this process further, since I don't want this to
be a he-said-he-said situ.

For example, is this something a bad mozconfig could do? I am certain home and
work aren't using the _same_ mozconfig, but they have the same author.

Comment 5

[Magnus Melin [:mkmelin]]

What am i missing?
This is my view: http://www.24.fi/magnus/test/snapshot1.png

You don't have any extensions running do you?

Comment 6

[Matthew Daniel]

(In reply to comment #5)
> What am i missing?
> This is my view: http://www.24.fi/magnus/test/snapshot1.png
> 
> You don't have any extensions running do you?

At least not that I'm aware of, since (as I stated in the initial report), I am
doing an `rm -rf $HOME/.thunderbird` for each test run.

Since your screenshot shows several e-mail accounts (I am only using TB for
RSS), do you think that matters? Maybe `mv $HOME/.thunderbird $HOME/.tbird-off`
and try it out?

Since TB is based on Zilla, I'll move my $HOME/.mozilla out of the way, too, for
completeness.

Comment 7

[Matthew Daniel]

I tried it with no .mozilla or .thunderbird and it behaves in the exact same
way. I'm going to create a new user to test with on my machine and that'll be
the best isolation I know how to provide without a specific request.

Comment 8

[Matthew Daniel]

I created a new user (which initializes its home directory to FC4 /etc/skel, but
it does NOT include mozilla or thunderbird settings). The good news is that it
didn't work for the new user, either (meaning its not a .rc problem). The bad
news is that it didn't work for a new user, either.

Here is my .mozconfig, located at the top of thunderbird/mozilla CVS working
directory. I don't know how to get to a "about:config" in TB, so if you have any
ideas for that, please advise.
=== BEGIN .mozconfig ===
. $topsrcdir/mail/config/mozconfig

export BUILD_OFFICIAL=1
export MOZILLA_OFFICIAL=1

mk_add_options BUILD_OFFICIAL=1
mk_add_options MOZILLA_OFFICIAL=1
mk_add_options MOZ_OBJDIR=@TOPSRCDIR@/obj-@CONFIG_GUESS@

ac_add_options --disable-debug
ac_add_options --enable-optimize="-Os -m64 -pipe"

ac_add_options --enable-xft
ac_add_options --disable-freetype2
ac_add_options --enable-default-toolkit=gtk2

#vim:ft=sh:
=== END .mozconfig ===

Comment 9

[Magnus Melin [:mkmelin]]

(In reply to comment #6)
> Since your screenshot shows several e-mail accounts (I am only using TB for
> RSS), do you think that matters? Maybe `mv $HOME/.thunderbird $HOME/.tbird-off`
> and try it out?

Well, I'd try adding an email account just to make sure... Would be quite odd
though.

Also, in the initial report you said 1.5b2, but still you talk about a CVS
version? If you aren't using the official build (from mozilla.org) do try it out
first, just in case your home made build is causing the problems.

Comment 10

[Matthew Daniel]

(In reply to comment #9)
> Also, in the initial report you said 1.5b2, but still you talk about a CVS

Yes, that's why the word "and" appeared in between those two versions. Maybe I
should have used mathematical set notation to say "inclusive." I am very
disappointed with how this bug has been handled. If you had simply followed my
bug report (blow away .thunderbird, subscribe to an RSS feed, click on an
article) you would have seen this bug _instantly_.

Instead the burden of proof has been on me, because I took the time out of my
life to try and make TB better.

For anyone who cares, I have the _EXACT_ steps to recreate this. Not that anyone
will follow them...

> Well, I'd try adding an email account just to make sure... Would be quite odd
> though.

That was the problem.

STEPS:
1. download thunderbird 1.5b2 from mozilla.org.
2. Start Thunderbird
3. Don't import anything
4. Next
5. RSS News & Blogs
6. Next
7. News & Blogs
8. Next
9. Finish
a. Click News & Blogs
b. Manage Subscriptions
c. Add
d. http://download.freshmeat.net/backend/fm-releases-unix.xml
e. Check Show article summary
f. OK
10. Click X to close that dialog (as there's no OK)
11. Click the + next to News & Blogs
12. Click on freshmeat.net
13. Click on the first article you see in the list. Sorry, I can't specify the
exact text for this part.
14. OBSERVE THERE ARE NO HEADERS
15. Click on the Write toolbar icon
16. Email account should already be chosen
17. Next
18. Next
19. type "fred" in the "Incoming Server" text field
1a. type "joe" in the "Outgoing Server" text field
1b. Next
1c. Next
1d. Next
1c. Finish
1d. Type "joe" in the To field
1e. Type "joe" in the Subject field
1f. Click the Save toolbar icon
20. Click X to close the Compose window
21. Click the + to expand Local Folders
22. Click on the Drafts folder
23. Click on the mail with the Subject "joe"
24. Observe there ARE headers 
25. Click on freshmeat.net feed
26. Click on the same article or a new one
27. OBSERVE THERE ARE HEADERS NOW
28. APOLOGIZE TO THE BUG WRITER FOR NOT FOLLOWING HIS ORIGINAL INSTRUCTIONS

Comment 11

[Magnus Melin [:mkmelin]]

-> new using
Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.8b5) Gecko/20051019 Thunderbird/1.5
ID:2005101906

With nothing but the one rss account in my profile i get this in the JS console:

Error: [Exception... "Component returned failure code: 0x80570016
(NS_ERROR_XPC_GS_RETURNED_FAILURE) [nsIJSCID.getService]"  nsresult: "0x80570016
(NS_ERROR_XPC_GS_RETURNED_FAILURE)"  location: "JS frame ::
chrome://messenger/content/msgHdrViewOverlay.js :: anonymous :: line 400"  data: no]
Source File: chrome://messenger/content/msgHdrViewOverlay.js
Line: 400

which would be
http://lxr.mozilla.org/mozilla1.8/source/mail/base/content/msgHdrViewOverlay.js#426

Seems there is no abook.mab set up yet. As a workaround, you can copy abook.mab
into the profile and headers will show up. Or simply set up a mail account;)

Comment 12

[Ville Pohjanheimo (inactive)]

Unfortunately these problems are not limited to RSS accounts, but also to IMAP accounts. If the first created account is an IMAP account, no headers are shown.

The JavaScript error I get with an IMAP account is identical to the one quoted in comment#11.

This was with TB 1.5 RC1 candidate (20051028) on Linux, but I saw the exact same behaviour on a Finnish 20051025 build on Windows.

Requesting blocking RC1 if that's still possible...

Comment 13

[Scott MacGregor]

too late for this release for something I ahven't been able to reproduce yet either. If I believed this was really wide spread we would have had lots of reports of this too.

Comment 14

[Ville Pohjanheimo (inactive)]

Ok. Hope nobody considers going over the steps to reproduce as bugspam... as I'll do that now (once more). Here's how to with a recent (en-US) branch TB:

1. Start TB with a new, clean profile
2. Create a valid IMAP-account
3. Select any message from the list of e-mails on the account
4. Observe no headers.

This works every time with my fastmail.fm IMAP-account.

Now, create an addressbook entry and select any message from the same list of e-mails and observe headers are visible and functioning again.

Comment 15

[Magnus Melin [:mkmelin]]

Yeah, I can easlyt reproduce for for IMAP too. Fresh profile needed. 

Regression window (the linux builds):
* 2005-09-30 working
* 2005-10-03 broken

Very likely fallout from bug 285474. 

Comment 16

[Magnus Melin [:mkmelin]]

*** Bug 314151 has been marked as a duplicate of this bug. ***

Comment 17

[Magnus Melin [:mkmelin]]

Linux here. Mac/win in the dupe. 

Comment 18

[Hiro]

Attachment: screen shot of Mail view with POP

Even only POP account reproduces.

Mac OS X 10.3.9
Tb 1.8branch build version 1.5 (20051028)

Comment 19

[Hiro]

Trunk build reproduces. 

Mac OS X 10.3.9
Tb trunk NB version 1.6a1 (20051029)

Comment 20

[Hiro]

Attachment: screen shot of message window

The message window is also similar.

version 1.6a1 (20051029)

Comment 21

[Ian Neal]

(In reply to comment #15)
> Yeah, I can easlyt reproduce for for IMAP too. Fresh profile needed. 
> 
> Regression window (the linux builds):
> * 2005-09-30 working
> * 2005-10-03 broken
> 
> Very likely fallout from bug 285474. 
> 
Yes, now that the correct variable is being used it is getting to this bit of the code which falls over if no address book exists to collect addresses to.

Comment 22

[Ian Neal]

Attachment: First pass at regression fix v0.1 (Checked into trunk)

This patch:
* Adds a try/catch
* Does some tidying of the code - less duplication hopefully
* From looking at the code, I think dontCollectAddress is now being used correctly as well

Those that can, please try this patch and report back here.

Comment 23

[Hiro]

(In reply to comment #22)
> Created an attachment (id=201300) [edit]
> First pass at regression fix v0.1

I tried this patch with RSS account and POP account.
The result is excellent.

Windows XP SP1
Tb 1.8branch build version 1.5 (20051025)

Comment 24

[Matthew Daniel]

(In reply to comment #22)
> Created an attachment (id=201300) [edit]
> First pass at regression fix v0.1

Confirmed successful fix for the RSS (I didn't try IMAP)
HEAD(20051030)+patch on Linux/x86

Thank you for looking into this.

Comment 25

[Scott MacGregor]

Comment on attachment 201300 [details] [diff] [review]
First pass at regression fix v0.1 (Checked into trunk)

I can easily reproduce this as well.

The fix looks fine.

Ian, I'm wondering if we should try for a 'safer' patch for the branch that just added the try/catch instead of tidying up the code and fixing the dontCollectAddress logic. What do you think?

Comment 26

[Scott MacGregor]

I ask because if we tried to take a super safe fix for the branch we'd have a very limited amount of time to test it. 

Comment 27

[Ian Neal]

Attachment: Branch version of fix v0.1

Just try/catch version of trunk patch for branch.

Comment 28

[Ian Neal]

Attachment: Branch version of fix v0.1 (non-whitespace changes) (Checked into branch)

As per previous patch except not showing whitespace changes.

Comment 29

[Scott MacGregor]

Comment on attachment 201685 [details] [diff] [review]
Branch version of fix v0.1 (non-whitespace changes)
(Checked into branch)

thanks Iann.

Lemme know if you want me to check this into the 1.8 branch.

Comment 30

[Ian Neal]

Comment on attachment 201685 [details] [diff] [review]
Branch version of fix v0.1 (non-whitespace changes)
(Checked into branch)

Checking in (branch)
msgHdrViewOverlay.js;
new revision: 1.56.2.4; previous revision: 1.56.2.3
done

Comment 31

[David :Bienvenu]

Comment on attachment 201300 [details] [diff] [review]
First pass at regression fix v0.1 (Checked into trunk)

probably don't need this +, but clearing out my queue.

Comment 32

[Ian Neal]

Comment on attachment 201300 [details] [diff] [review]
First pass at regression fix v0.1 (Checked into trunk)

Checking in (trunk)
msgHdrViewOverlay.js;
new revision: 1.61; previous revision: 1.60
done

Comment 33

[Magnus Melin [:mkmelin]]

*** Bug 315299 has been marked as a duplicate of this bug. ***