Thunderbird/StatusMeetings/2008-10-14

last meeting | index | next week »

Thunderbird Meeting Details

• Tuesday, Oct 14, 16:30 UTC (9:30am Pacific, 12:30pm Eastern)
• How to dial-in

Agenda

• Who's taking minutes? --> asuth
• Minute taking Schedule. Talk to davida for schedule changes/additions.

Action Items

• gozer to file bug re getting alpha bits signed (bug 459400)
• gozer to ping mrz about separate messaging crash-stats server
• dmose to do another round of updates on reviewers

Shredder Alpha 3

• Shredder Alpha 3
• Standard8 is the release driver
• wsmwk is the QA lead of Beta 1

Thunderbird 3 Beta 1

• bienvenu is release driver
• Proposed String freeze date: 2008-11-13
• Code Freeze date: 2008-11-18

Thunderbird 3

• Draft Thunderbird:Thunderbird3:DevRoadmap identifies major work items on the plate for Tb3 -- much of it needs scheduling, owning, breakdown.
  • folks interested in helping with front-end stuff, some simple, should ask in #maildev

Roundtable

Status Updates

davida

• Was in Quebec giving a talk at the CLLAP -- slides will go up today.
• Planning trip to France next week & Spain for MozCamp EU
• recruiting volunteers for various things

dmose

• lots of Gloda review work
• impromptu addressbook directional meeting
• booked FSOSS trip & prepared for talk
• driving work
• discussion about gloda-as-ext-vs-module
• gloda & calendar testing & bug reporting
• various personal organization/project work
• Thunderbird reviewer update work

jcranmer

• kill-rdf work, especially bug 457333
• Started work on a JS version of nsMsgDBFolder ([1])
• Released QAserver: [2]

Standard8

• Reviews & Driving
• Wrote some pages to guide/aid future releases:
  • Thunderbird:Doing_Releases
  • Thunderbird:Doing_Releases/Writing_Release_Notes
• Created new page for MailNews:Automation With Gristmill and filed some bugs
• Landed Growl integration patch bug 308552 and filed follow-up bugs.
• Autocomplete
  • Dropped old xpfe based local autocomplete code.
  • LDAP transfer still to do, WIP patch for sorting out connections, but still quite a way to go on that.
• Address Book
  • Fixed bug 459358 It is possible to attempt to create new cards/lists in read-only address books.
• Misc
  • Did some debugging on IMAP fake server hangs bug 437193
  • Wrote some patches and test cases for bug 326809 Modernize nsIMsgHeaderParser - |string| and |wstring| vs AString and AUTF8String

asuth

• initial calendar tab-unification with exptoolbar support landed bug 402365; dealt with some cleanup/fallout from that, one cleanup patch rotted, need to fix that.
• extensive gloda refactoring/reimplementation in terms of:
  • attribute representation and storage throughout the system. Previously, attribute providers returned a list of attributes pre-chewed for the database. This made it hard/impossible for providers to override attributes provided by earlier providers. Additionally, a lot of potentially confusing and inefficient happened by the scenes to expose the attributes on the object. Attributes are now set directly on the object as their representation objects, allowing subsequent attribute providers to override their predecessors.
  • attribute loading. Previously, attributes were loaded on-demand in a synchronous fashion from the database. We now just store the attributes in a JSON BLOB on the message row. This avoids the need to issue additional queries to get information for the messages in a big way. The extra storage cost is somewhat (entirely?) offset by the elimination of two fully covering indices from the database.
  • attribute provider / indexing execution model. Additional abstraction has been added to let attribute providers and key indexing logic take advantage of the generator-based execution model. This allows their code to exist as a single function that uses "yield" whenever it has to involve a (potentially) asynchronous process, including other indexing logic.
  • database loading / synchronous db access avoidance. SQLite is great but we simply cannot have synchronous reads (that cannot fail) from the main thread given our use of the async thread. (At least with our use of the FTS3 full-text search functionality which precludes the use of a shared connection cache.) Accordingly, the collection logic is being updated to (efficiently) load all referenced objects from a main collection in sub-collections before returning the results of the main collection. For example, if you are loading a message, it will make sure that all the identities are loaded (and in turn, all of the contacts associated with the identities) before providing you with the message.
    • This is actually necessary for the attribute changes, too. Because the attributes are no longer magic, we can't load the identities/contacts on demand.
    • This also nicely ties up some loose ends in terms of caching. Namely, by associating the sub-collections of identities/contacts with the collection of messages, we are ensuring that all of those contacts/identities will be reliably available from the collection manager to ensure our desired invariant that a gloda object representation in memory exists only once in memory.
  • database query mechanism is being cleaned up/normalized; the constraints using dubious variants of the original 'universal APV' (attribute, parameter, value) tuple are being done away with for more straight-forward cases. The goal is also to support query performance optimization by creating temporary tables from pieces of queries that can be re-used. The idea is that as the user adds constraints we can perform the query, so that by the time they decide on the final constraint, the database has already done most of the work it needs to do. This probably won't be completed in this refactoring right now, for time reasons. But I'm planning for it.
  • message indexing is losing much of its special-casing where generic paths exist.
• (note that all this gloda work is in progress, although I'm getting very close now)
• have been ticking dmose's extensive review action items off the list; the major refactoring above should fix most of the major issues (and scary errors dmose sees)

bienvenu

• Reviews and Driving
• Gloda testing and landing strategy discussions
• Cross-folder threading/grouping patch put up for review. I've also fixed some bugs with the patch and will put up a new patch once the initial review comments are in.
• Continued work in de-rdf repo. Synced with trunk, and fixing the known issues one by one. I hope to start the review process by the end of the week though I will need help with SeaMonkey testing.

gozer

• Build
  • Thunderbird 3 Alpha 3: DONE
  • Sunbird comm-central buildbots: ALL GREEN
  • Lightning comm-central buildbots: ALL GREEN
  • Leak/Bloat builds
    • One OS X builder is successfully reporting numbers (in MozillaTest)
    • Still need to plug-in TB's special profile setup (right now, running a vanilla start/stop cycle)
    • Still need to bring up the other 2 platforms
  • Turns out we are most likely hit by (hg bug 1320)
    • we are trying a workaround for this issue bug 459396

• L10N
  • Merged our changes back into l10n.mozilla.org
  • Figured out how to link into Simile (i.e. tb30 green locales)

• Misc
  • Got my phone stolen, might get a new #

clarkbw

• Out for the day, checking email
• Booked trip for EU_MozCamp_2008
• bugs I'm continuing to work on
  • bug 448288 — group actions collapsed threads
  • bug 452440 — message pane shouldn't be blank
• thunderbird search
  • actively working on search results page
• address book
  • pushed address book in tabs to the back of the queue bug 457270
  • worked on edit / view contact prototype (for post tb3)
  • need to work on listing display for AB and search results (for post tb3)

wsmwk

• 3.0a3 QA successfully completed - good turnout
• bug triage - profile bugs
• bugday

KaiRo

• bug 445141 build system cleanup: step 1 landed, more to do.
• bug 459431 another batch of m-c -> c-c sync: up for review
• bug 456373 build system target for source tarballs (can be used for any product): in review

rebron

• Continuing work with web site redesign. We're in the second round of creative review.
• Good progress with SpreadThunderbird. Targeting a soft launch of the new marketing web site next Thursday. We're meeting every Thursday at 1pm PDT.
• Continued work with market research and market segmentation.