"John Bandauer reports progress on XPConnect:
Non-success nsresult codes from native XPCOM methods now throw JS exceptions
when JS is calling native code
JS exceptions map to nsresult error codes when native code is calling JS.
The 'Component.interfaces' branch is working well enough to lazily reflect
IID objects by name or by canonical numeric string form as JS objects;
Helped fix enough bugs to get AppRunner on its feet.
We missed the browser BM2 milestone originally planned for last week, and
had to delay it until the end of next week.
Discovered that we had no written plans for dynamic menus. We will
be working on our plan some more starting next week to ensure that we haven't
missed any other requirements, and to update the task breakdowns and estimates
based on our more recent knowledge. If any groups have any dependencies
on XPToolkit that aren't listed in our requirements documents posted on
mozilla.org, please let us know immediately.
Finished (hopefully) MS-Windows implementation support for dynamic menus.
This was an unscheduled item from a new requirement to have dynamic menus
Added Mouse Wheel Tracker support for MS-Windows. This was an unscheduled
item, this functionality would be required for release, and as far as I
know, was not on anybody's schedule. I need to have the "view" issues reviewed
Cookbook for use of namespaces in the examples.
Language Spec for namespace information and examples. Added section
on XUL DOM extensions.
Spent a good deal more time than planned fixing numerous bugs, including
various platform parity, build bustage and regression defects.
Solaris Workshop compilation fixes are in, we're one step closer to being
able to use Purify on Unix.
Identified stability as a major concern for the LinuxDev people.
Getting some decent memory and profiling tools were listed as the next
objective for Unix Development. Note this isn't on anyone's schedule and
everybody is having trouble buying time to work on problems like these.
Changed the tree widget's expand and collapse mechanism to work the correct
Fixed numerous bugs in the XUL content model.
Added support for "style", "id", and "class" in XUL.
Had a successful carpool to rename member functions that were interfering
with AOLs embedding efforts (Travis Bogard asks that we emphasize
the experimental nature of their efforts).
Provided tests for the autoconf system to bring all the Unixes on line;
switched appropriate code over to using the result of those tests, rather
than hand checking for every compiler we know about.
Threw the switch in |nsCOMPtr| to actually make construction or assignment
from raw COM interface pointers illegal.
Presented our AppCores vs. Services issues to Warren with a proposed direction
that would allow us to completely fold AppCores under the ServiceMgr.
Simpler is better.
We've got to get a new process in place that ensures that AppRunner gets
stable and steadily increases in quality, while the tree remains open long
enough for new development.
Nobody owns or has scheduled time for 'middle-ground' problems such as
getting Purify working, profiling, etc., which has made stability much
more difficult to achieve and will make it harder to maintain.
General purpose glue to mate an arbitrary native dialog with the AOM is
seeming harder and harder. Hooking any _one_ dialog up by hand is
no problem. Still trying different solutions here.
Scott Collins plans to spend some more time onsite, around the end of the
month, no firm plans (or plane tickets) yet.
Troy writes in with this update:
"Kipp's on hold for feature work and is full-time fixing bugs. The most
notable bug fix this week was work on CSS box model calculations; we're
doing a much better job now.
Chris has also put feature work on tables in hold until his bug list is
I'm working in table pagination. We now repeat header and footer
elements when splitting a table across pages. We're also very strict now
about only allowing one header and one footer element (any additional
elements are treated like ordinary body row groups), and the header and
footer must become before any body row group elements. The W3C folks
will be clarifying the CSS2 spec to resolve exactly how this works.
Here's some extra status from Chris that applies to GFX:
Postscript Printing is mostly complete this with the addition of forms.
Added the capability to read and parse AFM
files to get font metrics on Linux. We are currently getting it to work
GFX rendered form controls can now be enabled in the viewer. The button,
checkbox, radio button are usable, but
not bug free."
Nisheeth Ranjan has this update on the status of XML in Mozilla:
"Expat, James Clark's parser (http://www.jclark.com/xml/expat.html),
got plugged into Gecko this week. The parser is checked into the
mozilla codebase in the directory mozilla/expat. XML well-formedness
errors now get reported inside Gecko if the expat parser is enabled.
Currently, the expat parser is not enabled by default. Once the
remaining bugs are worked out, hopefully by early next week, we will
start using expat for all XML documents.
Look out for Vidur Apparao's speech
(http://www.gca.org/conf/xtech99/Thu-Talks.htm#Gecko) at the XTech'
99 conference (http://www.gca.org/conf/xtech99/xtecindx.htm) in San
Jose next week. He'll be talking about XML display capabilities
inside Gecko and showing some really cool demos."
Here's Akkana's update on the status of the editor (composer):
"Some work toward implementing Get|Set|Remove Properties and SelectAll.
More work on InsertBreak; looking into converting consecutive multiple
breaks to a <p>.
Implemented OutputText and OutputHTML (currently in the Debug menu of
viewer). This is the routine which will be called by other users of
the editor, such as the mail/news compose window.
Work on selection iterators, which should clean up DeleteSelection and
fix some of the selection-related bugs. This is still in progress; if
deletion gets worse in the next few days, please bear with us because
it will improve in the long run.
Started work on getting Paste working.
Clarified the structure of nsI*Editor interfaces, and implemented the
new structure; see the newsgroup (netscape.public.mozilla.editor) for
Some work on defining the Rules system.
Phil Peterson has news regarding the Mail/News client:
Added directory /mozilla/mailnews/ to the Tinderbox builds on Win32 and
Linux, thanks to sar.
To see what we have so far (which isn't much), run the command line apprunner
-url resource:/res/samples/mailshell.xul from the dist/bin directory.
Parsing local mail folders works.
Displaying local mail messages works
Displaying the list of local messages in the thread pane works (using fake
data so far)
Some work on Reply and Forward commands.
Planned integration of mail/news with new security code (which of course
is not available to mozilla).
Plugged new netlib stream converters from ebina into libmime.
XUL files written for main 3 pane window, including menus and toolbars.
Worked with XPApps and netlib group on a plan for modal, blocking password
dialogs in the absence of Gecko reentrancy.
Unified component manager and service manager.
Some work on reading prefs, esp. for local mail path.
Planning to have some demonstrable mail functionality by 3/12
Need to plan milestones for post-3/12 releases, including IMAP, LDAP, prefs,
filtering, and other features.
Hopefully the MORK database will be up; we're counting on it for the 3/12
milestone to display real data in the thread pane.
Need to pass commands from XUL down through RDF to the mail/news data sources.
Need to resolve issues surrounding view-specific information needed to
dispatch commands (analogous to MSG_Pane from 4.x)
Need to find out how to re-root the thread pane and message pane based
on click handlers in the folder pane and thread pane, respectively"