[Chandler-dev] [Sum] Oct 9 - 15
Katie Capps Parlante
capps at osafoundation.org
Wed Nov 1 12:19:52 PST 2006
Build, Release and QA
*Checkpoint for Oct 9*
*Checkpoint report format*
Brian Kirsch asked about bugs listed in the checkpoint report that have
already been fixed. Heikki pointed out that the report has a section for
new bugs filed, even if they have been fixed. Heikki proposed changing
it to new valid bugs that are still open, and Brian agreed.
Given the amount of bugs we were finding and fixing, Philippe proposed a
few process changes for Alpha4. (1) Daily bug councils to triage bugs
more promptly. (2) Devs can check in not-yet-targeted bugs that pretty
obviously should go in alpha4, if the patch author and reviewer agree.
As of Thursday Oct 12, alpha4 had 13 blockers and didn't feel
particularly stable. We pushed out the release by a week (to Tuesday Oct
24th). We made a decision that we'd slip the date again until we had 2
"quiet" days without new blockers.
*In and Out Collections*
Brian Kirsch started implementing the fixes for In/Out collections, and
wondered how he could go forward without using filter collections. The
In collection membership would be based on to/cc fields, and the user
can change these fields at any time. He looked at using an observer, but
ran into problems wrt observing a list of emailAddresses (needs to
detect list membership changes as well as changes to value of an
PJE suggested we don't have items for emailAddresses, or not allow
changes to EmailAddress.emailAddress. Jeffrey agreed with not allowing
changes, but noted that emailAddress as Item will be useful for birefs
once we have contacts. Jeffrey asked if disabling changes to
emailAddress would solve the observer problem.
Morgen summarized a virtual meeting about sharing futures. The plan is
to change the way conflicts/merges are handled. The current strategy
rewinds changes to the time of the last sync, downloads the the server
changes, then merges by doing a refresh(). This approach turns out to be
slower than we'd like. Morgen will make a change to have the item
importing code check the repository history to detect conflicts. The
sharing framework code will pass a data structure describing local
changes during the relevant time period to the implementors of
importProcess(). Morgen has reviewed this strategy with implementors
(that was the virtual meeting).
Phillip finished a rough implementation of phase I of his EIM
implementation plan (primitive types, aliasing, conversion). He sent out
the doctest and asked for review and comments.
Brian Kirsch responded to a call for PyCon topics. He recommended a
standalone talk on EggTranslations, which we developed for Chandler and
might be useful in other contexts and fits well with an egg focus.
More information about the chandler-dev