Build, Release and QA
---------------------
*Weekly checkpoint*
Bear spins:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007827.html
Dan tests:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007840.html

*Logging SSL exchanges*
Heikki points out that network sniffers are nearly useless for debugging sharing or email problems over SSL. He gave instructions on how to print the exchange to stdout or a log file:
http://wiki.osafoundation.org/Journal/LogSSL20070227

*wxPython*
Robin updated wx to bring in the latest 2.8 branch changes:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007848.html

*Error exit*
Heikki explained that Chandler now exits with error code 1 if functional or performance tests fail, or if any script execution raises an exception. Making the behavior align with unit tests makes it easier to catch test failures from driving scripts.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007833.html

*Preview schedule*
Philippe gave an update on the preview schedule, which we changed to make more realistic:
- Feature complete moved out by 3 weeks
- Added 3 week performance tuning period
- Increased debug period to 8 full weeks
- Added precision to milestone definitions (agreement on what it means to hit each milestone)
Schedule milestones:
- Feature Freeze
  ... 3 weeks ...
- Performance Milestone
  ... 8 weeks ...
- Code Complete
  ... 2 weeks ...
- Launch
http://wiki.osafoundation.org/Projects/MilestonesDefinitionAndExitCriteria
http://wiki.osafoundation.org/Projects/ChandlerMilestoneSchedule
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007845.html

Dev
---
*Smoke tests for calendar changes, performance*
Hearing that John might do some performance work on the calendar UI, Jeffrey requested that care be taken if changes are made, as we don't have automated tests for all of the things we need to get right. Jeffrey listed out smoke tests he runs when making changes in the calendar. He also offered thoughts on optimizations, including:
- only redraw area of changed events
- avoid 2 redraws when creating new item (related to notification hints)
- only redraw in one calendar canvas (if changes only happen in one)
- fewer queries to repository for items in the current range (cache)
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007828.html

Andi mentioned an idea from IRC: events in range for the all-day and timed canvases could be cached on the UI event.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007829.html

*EIM: what attributes get shared?*
Morgen asked for a review of the domain model to make sure we had all of the appropriate fields to be shared in the EIM records. The records are listed/annotated in parcels/osaf/sharing/model.py. Adding attributes requires:
- adding fields to records (or creating records)
- implementing import/export translators
- agreeing with Cosmo team on serialization
http://tinyurl.com/26zfn8 (model.py)
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007831.html

*Deferred observers*
Andi implemented "observersDeferred".
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007832.html

*Dashboard*
A design list conversation about recurring events and the dashboard wandered in to the chandler-dev list. Points made by Grant and agreed to by Jeffrey: - adding/subtracting reminders in dashboard should apply to selected row/occurrence only - "occurrenceRemindable.userReminderTime = None" should do the right thing. If not, bug.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007834.html

*Byline*
Mimi noted that we don't really need a byline (beyond the updated date) if users aren't sharing or using email. Mimi is fine w/prompting the user for a name.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007841.html

*Proxied items*
Reid added a test menu item to force a conflict. Reid asked why he saw event handler code using "__getProxiedSelectedItems" and event updateUI methods using "__getSelectedItem". Grant explained that one only needs the proxy if one is going to change the item: updateUI code just looks at the item. If you are setting item attributes, you should use the proxy. It will record edit/update changes, and show the recurrence dialog where appropriate.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007836.html

*setStatusMessage*
Heikki objected to displaying an exception in the status bar. He argued that it can't be localized and is not understandable to end users. Perhaps the message should point the user to a log. Brian Kirsch pointed out that any ChandlerException (or descendent) can be localized. John believes most users won't even be able to find logs, it should be part of the "talk back" mechanism.
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007843.html

Meetings, Announcements
-----------------------
Platform team:
http://wiki.osafoundation.org/Journal/Platform20070306

QA IRC session, verifying resolved desktop bugs:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007838.html

Apps team:
http://wiki.osafoundation.org/Journal/AppsMeeting20070308

Desktop eng:
http://wiki.osafoundation.org/Journal/EngineeringMeetingNotes20070308

OSAF internships:
http://lists.osafoundation.org/pipermail/chandler-dev/2007-March/007849.html
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

Open Source Applications Foundation "chandler-dev" mailing list
http://lists.osafoundation.org/mailman/listinfo/chandler-dev

Reply via email to