Closed Bug 424237 Opened 17 years ago Closed 16 years ago

first-start bookmark migration from SeaMonkey/Netscape/Mozilla Suite doesn't work

Categories

(Firefox :: Migration, defect, P1)

defect

Tracking

()

RESOLVED FIXED
Firefox 3

People

(Reporter: wolfiR, Assigned: Gavin)

Details

Attachments

(1 file)

This is Firefox 3.0b4.

If I remove ~/.mozilla/firefox and start Firefox the first time it asks me if I want to import from Netscape or Mozilla 1.x.
I have a valid SeaMonkey profile and therefore chose to import from that option.
Firefox told me that it imported the stuff successfully but it seems that absolutely nothing (no bookmarks, passwords etc.) has been imported.
Still not working with 3.0b5.
Don't know if that should/would block FF3 but let others decide on that.
Severity: normal → major
Flags: blocking-firefox3?
Just tried again with 3.0b5 again and checked in more detail.
Passwords/signons and keys (NSS DBs) were imported correctly.
Also cookperm.txt, user.js, downloads.rdf, history has been imported.

What's not there are bookmarks in any form.
Summary: profile migration/import from SeaMonkey/Mozilla doesn't work → bookmark migration/import from SeaMonkey/Mozilla doesn't work
Reed/Dietrich: can we get a confirmation? This would be pretty bad, but I'm really surprised we haven't heard about it before.
Priority: -- → P2
I tested migration of bookmarks via the Library, and it worked mostly as expected: A folder entitled "From Netscape 6/7/Mozilla" was added to the bookmarks menu, and contained all of the bookmarks from Seamonkey. The only issue I saw was that bookmarks on the toolbar were not added directly to the Fx3 toolbar, instead into a subfolder of the imported bookmarks.

I'll try a "first install" migration next.
Confirmed. Migration from Safari worked fine. The relevant errors indicate that there's a failure trying to get the profile directory:

WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file /Users/dietrich/moz/trunk/mozilla/browser/components/migration/src/nsDogbertProfileMigrator.cpp, line 296
*** Failed to get string homePageMigrationPageTitle in bundle: chrome://branding/locale/brand.properties
WARNING: NS_ENSURE_SUCCESS(rv, rv) failed with result 0x80004005: file /Users/dietrich/moz/trunk/mozilla/toolkit/components/places/src/nsNavHistory.cpp, line 532
Flags: in-litmus?
OS: Linux → All
Hardware: PC → All
Summary: bookmark migration/import from SeaMonkey/Mozilla doesn't work → bookmark migration from SeaMonkey/Mozilla doesn't work
Target Milestone: --- → Firefox 3
Flags: blocking-firefox3? → blocking-firefox3+
Another thing I noticed is that simply all in seamonkey's prefs.js is copied over to Firefox (including all mail.* prefs and so on).
Shouldn't only the prefs in http://mxr.mozilla.org/seamonkey/source/browser/components/migration/src/nsSeamonkeyProfileMigrator.cpp#340
be migrated?
(In reply to comment #5)
> Confirmed. Migration from Safari worked fine. The relevant errors indicate that
> there's a failure trying to get the profile directory:

the dogbert error, I think that's expected, that's the ns 4.x migrator.  probably shouldn't be a warning though.
Assignee: nobody → gavin.sharp
The new bookmarks importer doesn't work without a profile, since it depends on the bookmark service. The 1.8 importer doesn't seem to be broken, despite having that same dependency, though - looking into why that is.
Status: NEW → ASSIGNED
Summary: bookmark migration from SeaMonkey/Mozilla doesn't work → first-start bookmark migration from SeaMonkey/Netscape/Mozilla Suite doesn't work
(In reply to comment #8)
> The 1.8 importer doesn't seem to be broken, despite
> having that same dependency, though - looking into why that is.

Turns out I was just crazy, and didn't notice the |ifdef #MOZ_PLACES| around all that branch code.

I did figure out why this was failing for SeaMonkey and not (for example) Opera, though - most other profile migrators call DoStartup() on the nsIProfileStartup they are passed, which sets the profile and notifies profile-do-change and profile-after-change observers. But just doing that at the beginning of nsSeamonkeyProfileMigrator::Migrate doesn't work, since the SeaMonkey profile migration is mostly just copying the SeaMonkey files into our profile, and if we DoStartup() before that, some of the services that listen for profile-*-change notifications have already created the files we're trying to copy over, which leads to NS_FILE_ALREADY_EXISTS errors.

Given bug 386678 I think we want to avoid just copying over bookmarks.html, so in addition to calling DoStartup() before trying to import bookmarks but after the other copying, I'm going to change the SeaMonkey migrator to call ImportBookmarksHTML() rather than CopyFile(), and have it set importBookmarksHTML = false like the Opera/IE/Safari migrators.
Attached patch patchSplinter Review
Attachment #317034 - Flags: review?(dietrich)
Priority: P2 → P1
Whiteboard: [has patch][needs review dietrich]
Comment on attachment 317034 [details] [diff] [review]
patch

r=me, thanks for digging into this.
Attachment #317034 - Flags: review?(dietrich) → review+
Comment on attachment 317034 [details] [diff] [review]
patch

Testing first-run migration isn't trivial and requires a new test harness of some sort, but I'll make sure we get good litmus coverage here.
Attachment #317034 - Flags: approval1.9?
Whiteboard: [has patch][needs review dietrich] → [has patch][needs approval]
Comment on attachment 317034 [details] [diff] [review]
patch

a1.9=beltzner
Attachment #317034 - Flags: approval1.9? → approval1.9+
Keywords: checkin-needed
Whiteboard: [has patch][needs approval]
mozilla/browser/components/migration/src/nsSeamonkeyProfileMigrator.cpp 	1.40 
Status: ASSIGNED → RESOLVED
Closed: 16 years ago
Keywords: checkin-needed
Resolution: --- → FIXED
Looks like this is already mostly covered in Litmus:
https://litmus.mozilla.org/show_test.cgi?searchType=by_category&product_id=1&branch_id=15&subgroup_id=777

The migration ones lack precision, as Marcia mentions at http://wiki.mozilla.org/MozillaQualityAssurance:Litmus_Test_Case_Maintenance#New_Firefox_3_Test_Cases_that_need_to_be_added_that_aren.27t_covered_by_assigned_areas , but they still cover this case I think. I tweaked the first-run migration tests to mention bug 386678, and would appreciate someone checking my edits.
Flags: in-litmus? → in-litmus+
You need to log in before you can comment on or make changes to this bug.

Attachment

General

Created:
Updated:
Size: