making XPConnect a reality.
We've even thrown another body at the problem (Mike McCabe) to make the
work go faster. Code to handle non-trivial C<->JS type conversions
is almost done. We are on the verge of having typelibs
being both written by the XPIDL compiler and read in and used by XPConnect
at runtime. Additionally, John Bandauer has created an nsIXPCScriptable
interface for reflection of XPCOM objects with property names that are
dynamically determined at runtime. Mike Shaver has begun work on
the IDL Author's Guide.
We're ramping up our efforts with the ECMA
standards organization to evolve the ECMAScript language for the upcoming
Waldemar has been doing a
great job of writing the ECMA 2 proposal.
As you read his proposal, keep in mind a few things: First, it's a proposal,
not a standard ! Second, even as a proposal, it's a work in progress
and we welcome input
with its object model; You won't find any references to document
objects, windows, forms, etc. The definition of browser objects falls
under the DOM specification.
Since JS2 is still a ways off, Norris
has been looking at simpler improvements that could be introduced in JS1.5,
such as the introduction of packages.
Checked in initial dialog support and posted a Dialog
Cookbook. Please give us feedback ASAP so we can rev this quickly.
Cleaned up the Toolbar Requirements
Anyone planning to use toolbars should review these to ensure that the
finished widget meets their needs.
Wrote and posted AOM
Nodes as the XPToolkit and AppCore/Service APIs, which clarifies the
preferred mechanism of communication between XPToolkit and application
services. It also clarifies the role of command nodes.
Reimplemented the Submit, Reset buttons to be GFX rendered and checked
Simplified the Grippy Pane Widget task by eliminating it as a separate
widget, making it a part of the pane splitter.
Checked in the rework of nsWebShellWindow and BrowserAppCore to better
support the creation of Dialogs and windows. Made the BrowserAppCore a
listener of network activity as a starting place for notifications for
enabling/disabling and start/stop of various browser components when a
page is loading.
Performed initial work of turning the AppCorManager into a service. More
investigation and understanding of the problem is needed, because it broke
the connect to JS.
Added "fixed" position for the IFRAME for the AppShell.html file. Now the
StatusBar "sticks" to the bottom of the window when it is resized.
Rewrote the tree frame construction code in order to merge with Karnaze's
new table construction code.
Worked with Chris Waterson to get the tree widget displaying mail and bookmarks.
Implemented the broadcaster mechanism in the RDF content nodes (for the
Platform parity, build and regression issues continue to take up way too
much time. Mac folks are wasting at least 10% of their time, and
here is what Chris McAfee alone had to do for Unix this week:
Finished checking in fixes to the Unix unit tests, this required some Solaris
work to keep Tinderbox happy. This fixes 6 or so PP bugs.
Solaris purify doesn't appear to be working properly with egcs, I will
need to go see if the native compiler can compile mozilla, this might be
the only way we get purify on any Unix platform.
Fixed libimg linking problem that was preventing non-imglib binaries from
Fixed build breakage all day on Tuesday.
Spent 2 days with the Escalation team getting FullCircle Talkback into
4.51 for HP-UX. This would have taken weeks for someone else to ramp
up and do, we do not have the headcount to make this happen. Nortel/BBN
escalations from last year might go more smoothly this year with HP/Talkback
under our Talkback belts.
Worked with Jan to update the Bugzilla component list.
Scott Collins has headed back to work at home in Michigan for a few weeks.
From Chris Blizzard:
"Resizing top level windows now works in the gtk front end. This has
been a major problem over the last few weeks and the layout people are
quite happy that it's been fixed."
|Grendel (Java Mail/News client)
|Submitted by Giao Nguyen <email@example.com>
Giao has this update for us:
"It was a slow week mostly with bug fixes rather than advances.
We fixed a lot of bugs with language sanity for moving things to Java
2. Grendel was patched so you can change the layouts of folder
display, message, and message body display. It was quite
thrilling. Some missing details from the addressbook code committed
last week were added to make sure everything compiled.
There're currently 4 (is that all???) files that are not
compiling. Hopefully, we can get them all compiling and working and
not just commenting out the entire thing.
- Edwin got us a few patches for language sanity. Chars not typed to
bytes sort of thing.
- Jeff is still hacking on addressbook issues. He's also trying to
line up some QA on the work.
- I'm fixing two out of the four files that are not compiling. They're
related to how grendel displays dialog boxes."
First, Scott MacGregor has news regarding the networking aspect of the Mail client:
"We made some more progress on porting our mail protocols over to the
mozilla world using the new networking model. SMTP is now online in
(mozilla/mailnews/compose). Part of the mailbox url code (mailbox://) is
up and running as well. So right now we have NNTP, POP3, SMTP and
mailbox protocol modules checked in under mailnews."
Brian Ostrom has news on Unix platform work...
"I've been working on getting the shunned UNIX platforms building.
Mostly without much success. Many of my UNIX boxes won't build
GLib and/or GTK+ (1.1.14), so I can't go any further until that
gets resolved. However, I have set up Tinderbox builds on HP-UX
10.20 and BSDI BSD/OS 3.1. Unfortunately, they still require a lot of work before they'll go
green (and when that happens, they'll become part of the standard
set of SeaMonkey Tinderboxes, in theory)."
Here's Akkana's update on the status of the editor (composer):
"The nsIEditor interface is being checked in, along with some examples
of how external consumers might use the interface outside of the
editor. We hope to have a document soon detailing this; when it's
available it will appear under http://www.mozilla.org/editor.
Code to implement a blinking caret has been checked in and is almost
ready to be turned on. We've been shifting around some of the backend
selection classes to make it easier for code outside of the layout
library to get access to the selection and to be able to iterate
through it. Selection via the keyboard has hit a brick wall, that of
relating the geometric position of a frame on the page to its position
(if any) in the document, and how to get from one frame to the one
geometrically next to it. This is a major issue which will need to be
solved with the help of layout people. See jfrancis' post in
netscape.public.mozilla.editor for more details.
The spec on the Transaction Manager has been updated, and is available now. The txmgr is a very general class, intended for use
outside the editor, and we're talking to other groups about how it can
be used anywhere that undo/redo is useful.
A table editing spec, describing what had been implemented for the old
mozilla classic code, which we plan to implement again in the new
codebase, has been published at
Work on dialog/UI specs continues. See the ongoing discussion on the
newsgroup, and the specs posted at
Troy writes in with this update:
a few bug fixes
regression code so we can do automated regression testing
XPCOM'ing of the nsIPresContext/nsIPresShell API's
this week: colapsing row groups, rows, col groups, cols
last week: map arbitrary html elements to table related elements
via display types
this week: win32 - print borders around form controls (using CSS attribute
selectors); Mac - similar toWin32 but not fully tested; linux - postscript images
last week: win32 - print state of form control widgets; Mac - similar
to Win32; linux - postscript printing excluding form controls
this week: gfx radio buttons and check boxes are limping along
last week: underlying mechanism for supporting either native or gfx
rendered form controls