349073 - address book comes up empty after running TB 1.5 and then running newer release

Status: RESOLVED WONTFIX

(Thunderbird :: Address Book, defect)

Description

[Scott MacGregor]

Every time I run a branch build and try to access the address book, the branch sets the following user pref:

ldap2.servers.pab.filename   pab.mab

And we end up looking for the PAB in that location instead of in abook.mab, so from that point forward my AB is effectively empty when using trunk and branch builds. 

I don't see any references in lxr to pab.mab on the branch, I can't figure out where it's coming up with this value. Very strange.

I started seeing this a couple days ago.

Comment 1

[Scott MacGregor]

adjusting the summary

Comment 2

[Scott MacGregor]

I feel like this should be caused by Bug 346880 but that was trunk only.

Or the fact that that pref is now listed in mailnews.js is causing it to get erased from user prefs when you run a trunk build. And that makes it so the branch build no longer sees the name of the personal address book file, so it makes up a new one...

Comment 3

[Scott MacGregor]

I think comment 2 is correct.

I found a backup copy of my prefs file and I had the following user pref string:

user_pref("ldap_2.servers.pab.filename", "abook.mab");

When the patch in 346880 landed on the trunk, the next time I ran a trunk build, this pref was cleared since it matched a default pref. Go back to 1.5 or tb2, and the pref is gone so it creates a new pref value called pab.mab and creates a new file for my personal address book.

Comment 4

[Scott MacGregor]

I still can't tell where the branch is getting pab.mab from when it writes this value back out to user prefs...

Comment 5

[Scott MacGregor]

David suggested I look in localstore.rdf as a possible spot to see where tb2 / 1.5 is getting pab.mab from and I found this entry:

  <RDF:Description RDF:about="moz-abmdbdirectory://pab.mab"
                   NC:open="true" />

Maybe that's part of it.

Comment 6

[Mark Banner (:standard8)]

(In reply to comment #4)
> I still can't tell where the branch is getting pab.mab from when it writes this
> value back out to user prefs...

I think its formulating pab.mab from the pref name - I believe the pref name of ldap_2.server.<name> gives <name>.mab assuming no existing files etc. As the branch should still be enforcing that correctly, its a little weird.

I can't debug it til later, but it'd be useful to know if your collected address book (history) is exhibiting the same problems.

I'm wondering if we should put the patch on bug 346880 on branch, though that will mean if users go from a later version to an earlier one they'll be broken, but typically that's only for testing anyway.

The only other way would be to find an alternate fix for bug 346880, but I think that'll be difficult without re-writing the address book back end.

Comment 7

[Scott MacGregor]

(In reply to comment #6)
> (In reply to comment #4)
> > I still can't tell where the branch is getting pab.mab from when it writes this
> > value back out to user prefs...
> 
> I think its formulating pab.mab from the pref name - I believe the pref name of
> ldap_2.server.<name> gives <name>.mab assuming no existing files etc. As the
> branch should still be enforcing that correctly, its a little weird.
> 
> I can't debug it til later, but it'd be useful to know if your collected
> address book (history) is exhibiting the same problems.
> 
> I'm wondering if we should put the patch on bug 346880 on branch, though that
> will mean if users go from a later version to an earlier one they'll be broken,
> but typically that's only for testing anyway.

Let's go ahead and do that for now. I'll approve the patch in 346880 since you are ok with this too. 

Comment 8

[Mark Banner (:standard8)]

I've just checked the patch from bug 346880 in on branch.

Comment 9

[Frank Wein [:mcsmurf]]

Mark/Scott: Can this bug here be resolved as fixed?

Comment 10

[Scott MacGregor]

Moving this off the blocking list, based on the beta feedback, I appear to be the only person that has run into this issue.

Comment 11

[Wayne Mery (:wsmwk)]

Frank in comment #9:
> Mark/Scott: Can this bug here be resolved as fixed?

Scott, gone for you since checkin of bug 346880?

Comment 12

[Scott MacGregor]

Wayne, i still see this when I move back and forth between 1.5.0.x and 2.0.0.x. But no one else every ran into it and I'm trained now to clear the pref name when my PAB disappars after running 1.5.0.x.

Since no one else ran into this, I think we can mark this wontfix

Comment 13

[Bill Gianopoulos [:WG9s]]

If you put:

user_pref("ldap_2.servers.pab.filename", "abook.mab");

into user.js rather than prefs.js, I am fairly certain it will not be removed if it matches the default.