|
Highlights
-
David Hyatt has come up with what looks like a possible breakthrough on
the widget/frame interface issue that has been holding up the XUL loader
implementation. Checkin is pending agreement with the Gecko team,
and final implementation, but he has widgets loading, and it looks like
will be able to provide this to other teams in the coming week.
Lowlights
-
We have been spinning our wheels and getting little or no traction.
We missed the XUL-> DOM sink milestone for M3, but hopefully only by a
weekend. Other items that were dependent on this are in day-for-day slip.
Accomplishments
-
Finally getting some traction on Mac/Linux platform parity, having identified
the key bugs to fix, assigned owners and platform assistance, and begun
debugging.
-
Posted a draft of the Application
Object Model (AOM) document.
-
Merged the architecture and AOM documents together, built a detailed outline.
-
Made a new XPCOM page
for mozilla & further updated nsCOMPtr
user manual.
-
Data Model Widgets
spec written, posted publicly, and updated.
-
Command Architecture
spec written, posted publicly, and updated.
-
Posted a strawman XPFE
Drag and Drop story, including some basic requirements.
Decisions
-
We thought we had resolved the approach to widgets and extending their
formatting in a meeting with Gecko folks on 12/21, but new resistance arose
last week. We need to kill this snake, and stop playing with it.
-
Same deal for the decision on using XUL vs RDF for widget resources, which
is still getting resistance from some RDF proponents. We still plan to
use XUL, but if we can't get the loader working in the next week then we'll
have to fall back on using RDF.
-
We will not be using HTML 4 tables to implement the treeview widget.
-
XPToolkit (evaughan) will own the progressLib code.
Issues
-
We are repeatedly having problems getting decisions communicated to and
accepted by other teams, avoiding having them reversed, and getting cooperation
once they are made. One example is the old decision to use XUL rather
than RDF, another is the 12/21 decision to use the frame interface rather
than the widget interface for widgets. XPToolkit, XPApps and others
have been held up due to our inability to get and keep these issues resolved.
We are working to get them resolved in a way that everyone can live with,
but it feels very much like 'us vs them', and with some teams we
are spending more time in disagreement than cooperation.
-
I don't think we are adequately engaging the net, or getting a good return
on what little effort we are making to do so. We seem to post decisions
more than issues/discussions, which prevents the net from potentially making
valuable early contributions. We aren't always seeing this type of help
when we do solicit it early, but we certainly won't if we aren't consistently
trying.
-
Platform parity effort does not have full support and high priority throughout
engineering. Currently it has mainly involved the same Mac/Linux platform
experts who have always kept those platforms running. That doesn't fit
the new development organization.
People
-
We have 6-7 new Mac developers signed up to contribute on mozilla.org.
-
Scott Collins will arrive onsite Wednesday, and will be available to help
resolve any lingering issues starting that afternoon through Feb 2.
Chris Waterson writes in with this RDF update:
- Got a demo mail data source up and running and into the tree control.
Current builds should show a set of demo accounts.
- Fixed some RDF parser bugs in the RDF/XML data source. Can "round trip"
(write back) modified RDF files now.
- Working on getting notifications & commands wired up to the tree and content
model.
- Address book data source work underway
Here's Akkana's update on the status of the editor (composer):
Most of the editor team, being non-windows developers, have been called
to help with the Platform Parity Push and have been spending their
efforts helping other groups in getting the Linux and Mac code up to
speed.
But meanwhile, our Windows contingent has been busy adding
transactions: Editor Mode (in the Tools menu) is now hooked up and
allows typing (though it doesn't yet take into account the location of
the selection, so text gets inserted at the wrong place). There are
some editor tests hooked up to special keys, so don't be surprised if
odd things happen when hitting keys like Insert, Tab or Control-S in
editor mode.
Wan-Teh's update on NSPR (Netscape Portable Runtime):
Larry Hardiman finished and checked in his PLEvent work
(mozilla/nsprpub/lib/ds/plevent.{c,h}). He is running NSPR
tests on the new Linux 2.2-pre4 kernel this week.
Srinivas and I checked in some compiler warning fixes
submitted by Kathy Brade.
We also fixed NSPR implicit initialization problems
(Bugzilla bugs #2227 and #2248) reported by Radha
Kulkarni and John McMullen.
Daniel Nunes has this update for us from the Build/Release group:
The main tinderbox page (SeaMonkey), now has Solaris columns (in
addition to windows32, macintosh, and redhat linux).
The nightly binaries of viewer also seem much better this week than
last, though the build/delivery/ftp-site push automation is still a
bit flakey.
You can expect this to improve.
Troy writes in with this update:
- Steve moved off tables and onto the editor. Chris Karnaze is now the
table guy
- Kipp was on vacation for most of the holiday period...
- Because of that I got lots of work done. :-) The only things of
interest really are that if the HTML element has a 'transparent'
background then it renders the background specified for the BODY. This
maps things display better. Scrolling is now better. The HTML element
is scrolled, not the BODY element. And margins on the BODY are handled
better.
Grendel (Java Mail/News) has been resurrected, and we have a forwarded message from Jeff Galyan giving the schedule over the next few weeks.
"Okay, so we know what we each are working on for the next few weeks:
Jeff: grendel.addressbook, grendel.composition, getting opening a mail folder
(either imap or local) working
Giao: menu building from xml, event handling, HTML display"
Previous Updates
|