I have nothing official to report. Good progress is being made on the XSLT processor itself. I am just waiting for the actual Mozilla integration to finish before I start doing an official status update. It probably doesn't make sense to do the status update, until people can see the XSLT processor running inside Mozilla (of course they can still use it in standalone mode). The integration is pretty close however. Peter Van der Beken has pretty much taken over for Tom Kneeland on the integration side of things, and he is making good progress. I'll send you an official status update once the integration is finished. This shouldn't take too much longer.

As far as using the standalone processor, it's in pretty good shape right now. Most of the XSLT 1.0 recommendation has been implemented. I just added support for simple sorting, and will hopefully have some I18N related code in there soon. The sorting just handles ASCII for the time being.

Marina Mechtcheriakova and William Brown have started to do conformance testing, and comparisons to the Java based XSLT processors that are available. I have received some bug fixes and changes back from them already and will have them integrated this week.

Summary of the Mozilla Japan Meeting and Party:

• There were about 115 people in attendance
• There were 24 volunteers who helped set up the meeting place with registration, banners, video, audio, etc. The organizing committee members all wore the Mozilla T-shirts from 1st and 2nd parties in SF. Thanks, Dawn! They loved these T-shirts.
• Those who registered to attend were each given a Mozilla Hack sticker provided by Dawn and a T-shirt provided by Laura and ICP. The Mozilla-Japan T-shirt was a great hit!
• The following media companies sent reporters to the conference/party.
• Japan IBM developerWorks Linux
• Impress Co. Mado no Mori (Windows Forest)
• MacWire
• Daily ASCII Linux
• ASCII DOS/V Issue Magazine
• Monthly ASCII Magazine
• C Magazine
• Asahi Newspaper
• When I proposed this meeting, I had 2 objectives.
• To gather together Mozilla volunteers in Japan and energize the movement further.
• To provide informative sessions to the Mozilla community and the Press in Japan
• These 2 objectives were definitely achieved with resounding success.
• The meeting was very relaxed and everyone seemed to have great fun -- I read a bunch of comments since the party ended and they all expressed the feeling that the party was great, the 3 sessions were very informative, and they now feel much more inclined to get even more involved in Mozilla activities, start doing some new work on it, and plan a set of long-range ideas for Mozilla Japan.
• The meeting opened with a phone call to Bob Jung who congratulated all of us and then read the names of Japanese volunteers who contributed greatly. People in attendance all applauded the people who stood up in turn to accept Bob's commendation. This really set the mood for the entire event. Gleeful, Joyous, and proud of the fact that Netscape and Mozilla.org seem to be aware of Japanese volunteers' activities.
• Session 1: Intro to Japanese Mozilla activities
• started with a fairly long intro of Mozilla and Japanese activities by Ryuzi Kambe. He did an excellent job of summarizing the movement in Japan. There was a demo of a Windows calculator which was built on Mozilla. This was quite a show by Ikemoto-san. He showed the calculator which comes with the Windows 98 and then the one he built with XUL and JS on Mozilla. They looked nearly identical and functioned identically as calculators.
• Then Dai-san and Koike-san demoed the Japanese Bugzilla. This aroused a lot of curiosity.
• Kazu Yamamoto and Ryoichi Furukawa then talked about Japanese Lang Pack project -- a bug round of applause for these 2 people from the floor as they all got introduced to Mozilla through their work.
• Tam-Neko, Money, and D MOJI talked about the Mozilla.org translation project. They asked for new volunteers to help -- they got several people to come forward at the party to help.
• Session 2: How to hack Mozilla -- case studies
• The 2nd session opened with a phone call from Dawn Endico. We had pictures of the Pacific Ocean Dawn put up on her web site. She expressed the appreciation on behalf the Mozilla.org staff and encouraged Japanese volunteers to keep on doing the excellent work.
• I began the session describing two objectives I had for this session. One, how one can overcome the initial fear of looking at a huge code base and how one should begin. Two, how did the 2 invited panelists approach their code work.
• Kato-san began his speech with an overall i18n architecture explaining the relationships among NGLayout, XPCOM, XP ToolKits, and OS. He then identified where recent i18n breakdowns have been occuring in M15 and also where to look when multi-byte characters don't display as expected. I heard from many participants later that Kato did a fantastic job of conveying what hacking the Mozilla code feels like. Kato-san explained along the way how to build and start looking at codes.
• Katakai-san then explained recent XIM checkins in detail explaining how XIM support is dealt with in Mozilla and its relationship to Gtk. He was thorough and clear. People really appreciated his demo of the new XIM features on his Solaris 2.7 on his laptop.
• The panelists then accepted questions from the floor. Interestingly there were many questions about Gecko -- there were at least 5 people who are working on building an application on Gecko. Even some Mac people are doing it.
• Session 3: Mozilla present and future projects
• Retrospective by Yoshioka-san
• Dawn Endico and Bob Jung's wish list for Japanese Mozilla community -- Kat Momoi
• Internationalizing
• Open Panel sessions: Yoshioka, Momoi, Kato, Katakai, and Egota
• These panelists presented what the consider to be future ideas for Mozilla Japan community.
• Kato mentioned that i18n bugs need to be fixed -- he referred to a bunch of regressions which occurred during M15.
• Katakai-san mentioned further ideas for improving XIM.
• etc.
• The party: 4:30 - 6pm
• People mingled and talked. Lively discussions, contacts, people volunteering for projects. When the party ended, there were still over 50 people left.
• 30 people or so went to the 2nd party. (till about 9pm)
• 15 or so people went to the 3rd party (till midnight?)
• etc.

Mail messages that have been flying by for the past week indicate that there are now many more people involved in Mozilla Japan activities. Bugzilla Japan is about to come online officially, M15 Japanese Language Project is now under way, etc.
Here are some ideas for Mozilla.org-US to help out Japanese volunteers:

• Integration of Bugzilla-Japan at some appropriate point in future
• (Possibly) hosting Mozilla.org translation project pages (into Japanese)
• Helping with organizing for participation in O'Reilley's Mozilla tracks.
• Helping plan Mozilla.party.jp #2 next year by sending appropriate developer/reps.
• Increased trans-national activities -- Unicode Conference in San Jose this September should have a few people from Asia joining NS i18n/l10n members.
• Possible Mozilla international conference inviting volunteers from around the world and promote cross-national cooperation.
• etc.

The Linux GUI installer based on the XPInstall engine is now available with the daily builds.

Requires:
=========
(These should be present by default with any RH Linux installation.)
Local "zip" client to be on your path.
Local "ftp" client to be on your path.
GTK libs to be installed.

For developers:
===============
Please remember to keep the mozilla/xpinstall/packager/packages-unix manifest up to date as you'll have been doing for the counterpart packages-mac and packages-win manifests.

You can build your own installer (say to deliver branch builds to QA) in your local tree using:

> cd <your tree top>/mozilla/xpinstall/packager/unix/
> perl deliver.pl

You will find a self-contained installer delivered to:

<your tree top>/mozilla/installer/sea/mozilla-installer.tar.gz

I would like to thank Jon Granrose for integrating this into the release process rather smoothly.

Accomplishments
 

Mike Pinkerton reworked the clipboard api to handle the concept of multiple clipboards, specifically the idea of a special selection clipboard for X, in support of autocopy on selection for middle-button paste. (pinkerton)

 

Daniel Matejka implemented centering support for windows. (danm).

 

Eric Vaughan got splitters working in grids, but the checkin had to be backed out due to exposed problems on X.  (evaughan):

 

Stuart Parmenter (pavlov):

Fixed 35594 Numerous drawing problems on Linux, which should enable EVaughan's box grid code to be put back in.

Rewrote portions of Unix clipboard code to add support for autocopy  on selection for middle-button paste.

David Hyatt  (hyatt):

Fixed focus regressions.

Improved updating of :focus state.  Behaves correctly in most cases now.

Solidified XBL doc.  Troy has submitted it to the BECSS group.

The XPToolkit team resolved 57 bugs in the last week, fixing 23 of  these, including 5 new features and 5 PDT+.  For details, see our resolved bug list.

 

 

Priorities
 
 

pinkerton:

Make sure tooltips are usable for XPApps developers ( 22511 ?)

11586  XP Menus need to be scrollable

saari:

Dist chome directory restructuring for skinnability.

Look into mailnews focus issues on linux.

hyatt:

Help danm and saari with directory reorg of chrome structure

Skins bugs

evaughan:

Checkin grid widget, again.

Implement spans in grids.

pavlov:

 9668 XP cursor support.

danm:

Dist chome directory restructuring for skinnability.

Evaluate and check in helpful Robert O'Callahan's latest patches, which have been waiting for a couple of weeks.  He's been very understanding.

Tracking:

Durations of remaining XPToolkit feature work for PR2:

Last week:

DanM: 1 d
EVaughan: 6 d
Hyatt: 3+ d
Pavlov: 6 d
Pinkerton: 10 d
Saari: 0 d
Total: 21 d

Current:

DanM:  1 d
EVaughan: 6 d
Hyatt: 2.5+ d
Pavlov: 4 d
Pinkerton: 2 d
Saari: 1 d
Total: 16.5

Estimated feature work completion dates:

DanM:  4/18
EVaughan: 5/12
Hyatt: 4/21
Pavlov: 4/28
Pinkerton: 4/21
Saari:  4/18

Mail/News Status

• Beta2 Feature List
• Estimated Days to Beta2 Completion
  • Rich 10 (15)
  • Seth ? (10) [no status, Seth's on vacation]
  • ScottiP 10 (10)
    
  • MScott 16 (20)
  • Jean-Francois 14 (15) [added 7 for features not counted last week]
  • Bienvenu 15 (20+) [added search backend]
  • Candice 11 (20)
  • Jeff 25+ (20) [looking for update, don't have data yet]
  • Alec 15 (20)
•    Bugs closed last week
• Save all attachments basic feature checked in
• Bugs for this week
• 34962, 34570
• 33156, 17001
•  Bugzilla bug 16241 -- [FEATURE] Preferences Return Receipts item lists are NOT enabled  Bugzilla bug 17100 -- [FEATURE] limit_message_size isn't working (truncate message)
• 10875, 14106
• 11366 [FEATURE] Print Address Book Card
  32297 [FEATURE] D&D from Address Book to the Compose window
• 16354: [Feature] Autocomplete for mail compose
• 28898 - per-identity smtp picker
  17484 - making progress on the filter datasource
  
• Issues
• AB Sync depends on AOL server people responding to our request
• AOL mail server wierdness prevents turning on AOL psuedo mailbox view
  
• Debugging build tree regressions in general are taking a lot of our time
• Time lost to regression analysis of 35384, 35362, 22858, 33267 (not caused by profile manager changes)
  
• Autocomplete widget has dependency on bugs 35687 and 35690
• Bug  16244 -- need service to complete operation before shutdown
• JAR packaging demo ready this week (bug 18433)
  

Shrimp

• Estimated Days to Beta2 Feature Complete
• Gayatri 5 (15) [footprint work is Brendan's now]
• Writing proposal for DLL work (bug 17031)
  

Profile Manager

• Estimated Days to Beta2 Feature Complete
• Bhuvan 7 (10)
• Varada (10)
  
• Last Week
  
• Sent out profile migration proposal
  
• This Week
• build profile migration task list and schedule
• 34385 signed scripts for activation
• 10953 expose activation API to rest of app
  

CCK

• Last week
• closed some regression bugs
• This week
• handle any new regressions, no other work
  

Highlights

• Now with Gordon's status!
• Necko turns exactly one year old today! (Thanks to Andreas for pointing this out.)
• Cool job by Brian Ryner (Mr. Finger) on some docs for writing necko protocol plugins:
• http://www.mozilla.org/docs/netlib/new-handler.html
• Cool new Necko Status heading.

Engineering Status

• ruslan
• Fixed:
• 33762 Cache security problems (memory cache now caches security info)
• 34710 Need to make an exclude-list for servers which do wrong things with http/1.1 downloadable (note: the exclusion list is modifable yet - we want probably make it downloadable dynamically, but it's an RFE then; I downgraded the it to P5)
• 35024 Need an API on the socket transport service to tell if any connections are active
• 31957 crash opening popup windows & 35508 HTTP needs to fire OnStart's in all cases (implemented, but disabled till Travis fixes the webshell; the webshell doesn't like OnStarts in failure cases)
• 34547 Need to be able to configure nsSocketTimeout
• 34679 Need to add setters/getters for timeout
• 33952 Clicking on links or buttons sometimes does nothing
• 35787 Visiting expedia urls gives me http1.1 data in the content body!
• gagan
• Fixed bug 28950 nsHTTPHeaderEntry had nsString objects instead of nsCString.
• Fixed bug 32813 stand alone test for url parsing.
• Resolved bugs- 33537, 34648
• AsyncWrite changes- I have some asyncWrite changes for HTTP (bug 32560) but after talking with ruslan I may not check those in and clean up the HTTP api more.
• davidm
• mime data base parser written and tested on mac.
• Fixed a couple of bugs in the nsMIMeInfo implimentation and add some routines to help matt out
• Tracked down a bug in cache prefs not getting initialized
• Got nsStreamAsFile up and running for http. Still need to hook up file channel (don't know if I also have to do chrome/resource)
• andreas.otte@primus-online.de
• [35736 url encoding - character encoded when it shouldn't] fix checked in
• [34863 Local Harddisks not found if I tried with C: without \] fix checked in
• [33868 URL encoding for Basic Realm authentication doesn't work] fix checked in
• [34966 fragment id on root-documents without / misinterpreted] fix checked in
• [20891 Assertion when displaying message] fix checked in
• [32997 file URLs that include a hostname don't work] working on parsing UNC-filepaths for XP_PC only
• [33935 Open File on this test case causes crash] working on considering "/abc/def.ghi" as a relative url for XP_PC only since it doesn't start with a drive letter
• warren
• Landed second round of necko API changes:
• nsIFileSystem replaced by nsIStreamIO and subclass nsIFileIO (gives ability to get at I/O streams to nsIFile objects)
• nsIInputStreamChannel subsumed by nsIStreamIOChannel
• nsFileTransport is now completely generic -- given an nsIStreamIO (blocking I/O operations) it turns it into a non-blocking nsIChannel interface (should probably be called nsStreamIOTransport at this point)
• Jar protocol should now be in shape for our chrome/skins needs
• Concerned about finding someone to make the jar packaging work for the Mac build system
• valeski
• No status this week (a first!).
• gordon
• investigated bug 33821 "cnn.com crash in nsDNSLookup::CompletedLookup"
• this is actually a due to trying to Remove() a hashtable entry while we are in the middle of Enumerate()
• will either change nsSupportsHashTable to allow the use of the PLHashTable flags, or just use PLHashTable directly
• tracked down and fixed  Bugzilla bug 35031 "host resolving stops after periods of usage" with Ruslan.
• continued tracking down leakage of nsIOService with Patrick.
• Frank Tang has checked in a fix for one of the sources of leaks (nsURLProperties)
• CSSStyleSheetInner is the one leak remaining that prevents the nsIOService from properly shutting down on Mac
• triaged M16 & M17 bugs
• my M16 buglist looks manageable.  I'm concerned about the M17 list, however.

Schedule information

• No, I don't have a rolled up schedule yet, but here's the query of our beta2 bugs, most of which have duration information in the status whiteboard now:
• Bug List (65 bugs total)

Highlights:

• The team focused on beta2 features and M16 bugs.
• Simon enabled selection of text widgets from JavaScript and investigated the editing of pages having forms controls.
• Mike enabled autoCopy of selection to clipboard, and improved table cell selection.
• Akkana checked in code for the Composer View Source page and for the color customization bug.
• Joe got IME back in order.
• Charley implement "Show All Tags" mode and made big progess on property dialogs.
• Kathy has been working on drag and drop, and improving the image dialog.
• ? editor bugs were resolved this past week
• 110 open M16 bugs, 31 of those are feature work bugs

Lowlights:

• There are currently 241 open Composer bugs (M16-M30)
• Beth is sick (but will be back tomorrow)

Priorities for M16:

We used to have a "goals for this week" section here.  But it changed to "Priorities for M16".  So I left it alone.  A bunch of these are already done.

• Kin:
• find/replace
• spellcheck
• caret placement
• drag select
• Kathy:
• Composer preference panels
• rewrite image dialog
• drag-n-drop
• Joe:
• list transformation (changing list type)
• remove all styles
• bigger/smaller font selection
• tab support
• block operation
• IME support
• Simon:
• provide selection from JS
• command updating (caret and toolbar)
• hook up context menus
• disable unsupported elements
• split editor controller
• Mike:
• click selection
• drag select (set capture)
• select before/after table
• table cell selection
• selection thresholds
• page up/down
• Charley:
• rewrite all of the Composer dialogs
• table edit and selection
• row and column select and edit
• provide unknown element widget
• finalize menus
• Akkana:
• rewrite event system
• help debug several mail compose bugs
• rewrap in mail
• double-click word select
• disable meta refresh
• doctype support
• color customization

Accomplishments:

• Kin
• on sabbatical, back this week!  Woohoo!
• Kathy
• Made some progress on drag and drop
• Worked on changing Composer preference panels to not have any html tags
• Checked in Image dialog with all html tags removed
• Joe
• 3 days of IME regression fixing.  Bugs 34908, 34909, 35399, 35637, 23855, 33847.
• finished up "tab in table" implementation
• did "tab in list" implementation
• fixed bug 33946: inline props not working over cell selections
• began fixing problems with the block operations (ongoing).
• Simon
• Checked in changes to allow the selection of text widgets to be

modified from JavaScript (bug 33967)
• Fixed the select() method on text inputs and textarea.
• Fixed a build problem with the chrome/security directory
• Spent some time looking at how to handle form controls in editor
• (bug 25161). No clear solution yet.
• Mike
• Talked with travis some more about embedding.
• Have fix for 13277 in tree for home/end in tables
• Have fix for 28331 select all scrolling page.
• Have fix for 32708 fix for extra notifications flying around this will also help fix 7787 fix for selection copying as you scroll.
• Fix for 32900 Scrollbars selecting text as you scroll.
• Fixed 33731 ctrlright+shift selecting to the left. doh!
• fixed 26589 as much as reasonable until ender-lite
• Working on finding who I can talk to about nsIDocShell so I can have JSabled api for getting selection. (simon is doing this now.)
• bug fixes listed above have been checked in.
• Mouse dragging now copies to clipboard.(7787)
• Table cells now show their selection by setting their background color.
• Table cells now highlight with background of selection color from nsLookAndFeel.
• Sped up auto copy by adding another parameter called "aReason" to the notifications to let people know if its from a mouse movement.
• Fixed a bunch of other bugs resolved some from the fixes I did above
• Charley
1. Bugs fixed:
1. 35207: Make setting attributes in property dialogs use transactions
2. 35380: Remove links does not remove named anchors
3. 28498: Implement "Show All Tags" mode all done except for using new images from Beth/German
4. 28152: Implement "Recheck" button for Spell Check dialog and notification when done.
5. 34526: Replace editor's message dialog with global "Alert" dialog
6. 35885, 35785: Fixed null pointer bugs in xul layout causing crashes, such as launching Image Dialog.
7. 34894: Added "logical" HTML text styles (em, strong, etc)  to Format | Text Styles menu.
8. 20938: Added "Blank Page to Edit" to Browser, Mail, Address Book, and Message Compose modules "New" submenu.
2. Skins dialog rewrite (Main bug is 19565. Related bug fixed: 19565):
1. Most of the property dialogs are done: HLine, Link, Named Anchor, List, Color and Background, Image, Insert Source, Insert Table, and Spell Checking. Grids used when needed. Removed inline style, replaced "onclick" with "oncommand", etc.
2. Helper JS methods added to support <menulist> instead of <html:select> for "combobox" and to use <tree> instead of <html:select> for lists: ClearMenulist(), AppendStringToMenulist(), AppendStringToMenulistById(), ClearTreelist(), AppendStringToTreelist(), AppendStringToTreelistById(), and InitPixelOrPercentMenulist(). See EdDialogCommon.js for details.
3. CSS classes added for proper margins and padding: e.g.: "dialog" for <text>, "narrow" for <textfield>, others for minimum width and wide left margings. Using these, the dialog display is surprisingly polished (widgets line up, padding, and margins are getting better)
• Akkana
• 29794: investigated disabling meta-refresh in the editor.  Turns out it's something that the docshell is supposed to be able to do.  Added code in the editor to call the docshell method (which doesn't work) and passed the bug along to Travis, who agreed to look at it to see why the backend doesn't work.
• Renamed libender to libeditor on linux and windows.
• 35814: tracked Pavlov's change to the clipboard APIs to distinguish the global vs. selection clipboards (Unix CLIPBOARD vs. PRIMARY selections) by changing the editor Paste() APIs to take an argument for clipboard type.
• 15275: Fixed a regression in middle-mousing on a link that showed up in navigator.js when the new clipboard API was checked in.
• 28401: Cleaned up the key event handling code in nsGfxTextControlFrame.  Mike and Hyatt's changes may make further changes in this code unnecessary, but I checked in the changes just in case.
• Fixed the editor's view source dialog.
• 33753: Found a fix for our extra doctype tag in selections, but haven't checked in the fix yet.
• Checked in work from last week on 34704: Make nsXPLookAndFeel colors as well as metrics customizable.
• IRC conference with Intel re Weight Watchers for the editor
• Helped Harish debug a problem in his DTD rewrite.

---

Composer Engineering Beta 2 Feature Schedule

I didn't update this chart since I know that some stuff got work on out-of-order, and I don't know precisely how Beth would reprioritize the remaining tasks.

Please note -- the dates may change for the estimated fix date and the order in which the bugs get resolved may also change. The goal is to have, by the end of M16, all of the listed items in for beta2.

Tasks slotted for M16:
 

	10 April	17 April	24 April	1 May	8 May	15 May
Akkana	rewrite event system (2 weeks) 21 APR 28401	cont. with event system rewrite	help debug mail bugs (1 week) 28 APR 34941, 32336, 30376, 29653, 31906	general cleanup (on going) 6276 - rewrap in mail 16352 - double-click 29794 - meta refresh 33753 - doctype 34704 - color customization
Kathy	Composer preference panels (4 days) 13 APR 33293, 23593, 27840   image dialog rewrite (due to XUL changes) (2 days) 17 APR	cont. image dialog   drag-n-drop (4 weeks) 12 MAY 6255	cont. drag-n-drop	cont. drag-n-drop	cont. drag-n-drop
Mike	selection: click (3 days) 12 APR 16203, 16311, 27528, 32807   drag selection scrolling (set capture) (3 days) 17 APR 3836, 32028, 25131, 30609	cont. with drag selection   select before/after table (4 days) 21 APR	table cell selection (4 days) 27 APR 6270, 22751, 26831, 29785   selection thresholds (range of selection) (1 week) 4 MAY 15282, 25930, 33770	cont. with selection thresholds   pg. up/pg. dwn  (1 week) 12 MAY 4302	cont. with pg. up/pg. dwn
Charley	rewrite Composer dialogs (due to XUL changes) (3 weeks) 28 APR 19565, 34113, 34526	cont. rewrite dialogs	cont. rewrite dialogs	table edit   - edit/select rows/columns   - copy/paste cells (2 weeks) 12 MAY 26833, 20973, 26833, 26871	cont. table edit	unknown element widget (2 days) 16 MAY 22262, 26286, 28498, 34519   menus - finalize (2 days) 18 MAY 14342
Joe	list transformation (1 week) 14 APR 13513, 14753, 28456, 28716, 32733	remove all styles (2 days) 18 APR 26163, 27266   bigger/smaller (2 days) 20 APR   tabs   - lists   - tables (3 days) 25 APR 10929, 21290	cont. tabs   block operations (1 week) 2 MAY 22331	cont. block operations   IME support issues (on going) 23855	Composer style rules (4 weeks) NOTE: dropped
Simon	selection from JavaScript (editable selection placement) (1 week) 14 APR 33967	command updating:   - caret   - toolbar (1 week) 21 APR 34503	context menus (1 week) 28 APR 14662, 24686	disable unsupported elements (forms,frames,animated gifs, etc.) (1 week) 5 MAY 14768, 25161, 29162	split editor controller (need separated for text widgets) (2 days) 10 MAY 34504
Kin	on sabbatical	find/replace:   - case sensitive   - wrapping   spellcheck:   - wrapping   - case sensitive   - language selection   caret   - placement   - behavior   drag select - scroll text below window

Mozilla on BeOS is NOT WORKING YET! (forget M8 or M11). If you're looking for a binary, please check back in the future (best guess: 1 to 2 months). The Bezilla team has been waiting the past few months to recieve BeOS R5. The R4.5 compiler contained an error and was unable to compile Mozilla. Needless to say, this all but halted development. Now that Be has released R5, and the complier works, the Bezilla team is back at work, making up for lost time.

Richard Hess has started working on other projects as well, so he will be less involved but will still monitor the progress and help from time to time. Duncan Wilcox is busy working on Be Inc's related project. This means he is not able to spend as much time on Mozilla as we would hope. While we do need Duncan a lot, we also need to get new developers on board very quickly.

At the moment I'm trying to get this project up to speed, with the help of Duncan and Matthew. We will try to update this page frequently while helping new developers, such as Justin Morey on the netscape.public.mozilla.beos newsgroup, get up to speed and join our team. In order to get this project back on track, I've set some tentative goals:

a.. [Done] April 15th, 2000 - Ensure that M16 builds successfully (get you an executable).
b.. [In progress] April 25th, 2000 - Ensure that build instructions on this page work correctly.
c.. [?] April 29th, 2000 - Have all old patches from Duncan and Matthew checked into the tree.
d.. [?] May 15th, 2000 - Ensure that M16 builds a working local browser for Be - but without network support.

If you compile Linux mozilla with NS_TRACE_MALLOC defined in your environment (you need to compile this way only under xpcom/base and xpfe/bootstrap), then you can run mozilla with a --trace-malloc logfile option.  The logfile argument may name a file that's created, say /tmp/big.tm.  It can get large, but it's fairly compressed: a startup to first browser window trace is just under 6MB, in my (slightly out of date) build.

If you then run bloatblame with logfile as standard input, it will produce an HTML file containing two graphs, one linking libraries, the other linking "components" or "classes".  The graph nodes are ranked from largest to smallest according to total bytes allocated by calls into the library or component that ultimately reach malloc, calloc, realloc, or operator new.  For the example above, the invocation would be ./bloatblame < /tmp/big.tm > /tmp/big.html.

If you give bloatblame a filename as argument, it will write its HTML there, rather than on its standard output.  The -d option causes bloatblame to sort by Direct (or Self) allocation sums, not by Total (Self plus all libs or components that were called on the way to malloc or another allocator).  The -m min option suppresses graph nodes with totals less than min.

You may combine the mozilla run with bloatblame's execution via

# ./mozilla --trace-malloc "|./bloatblame -m 10000 /tmp/big.html"

(provided LD_LIBRARY_PATH=. or better).  If the --trace-malloc argument begins with a pipe character, mozilla pipes, forks, and execs the command specified by the rest of the argument.  The idea is to enable other tools, including realtime plotting programs, to be written for the "XPCOM TraceMalloc Log" file format specified by the magic "number" at the start of each .tm file.

See http://www.pavlov.net/mozilla/brendan/big3.html for a startup/profile-selection/first-window trace fed through bloatblame.  Click on the Fan-in and Fan-out links to follow the trail of malloc-blame through a maze of libraries and components.  Note that a similar game could be played with methods and functions, although that graph would be quite a bit larger.

Followups to mozilla.xpcom; suggestions welcome.

Highlights
 

Mike Pinkerton  (pinkerton):

Made tooltips fully functional again.

Can drag links from w/in content area (thanks to Kevin Puetz, puetzk@iastate.edu)

XUL popups now stay on screen when win32 taskbar is not at bottom of screen.

Daniel Matejka and Chris Saari completed their dist chrome directory restructuring, a large,  important part of the skinnability work.  (danm, chriss).

 

Eric Vaughan (evaughan):

Relanded box grid fixes that were backed out last week.

Added StyleChange coalescing that should dramatically improve performance.

Fixed some Gfx table horkage.

Stuart Parmenter (pavlov):

Fixed windows file picker to support modeGetFolder when using the new file picker interface.  This was keeping people from being able to change profile directories on windows. (bug #34071)

Fixed feature bug #13161.  Implemented nsClipboard::HasDataMatchingFlavors on unix so that editor, etc can ask the clipboard if there is anything on the clipboard.

Got copying multibyte text in to and out of mozilla on unix

working.  You can now copy and paste Japanese text in to
mozilla!  This fixes bug #33850.  This makes the clipboard on unix much more robust and makes Mozilla one of a very few apps on Unix who properly support the clipboard.

David Hyatt  (hyatt):

Tree performance enhancements, including making tree scrolling faster in mail/news.

Massaged the CSS and  XUL files to conform to the new directory org.

The XPToolkit team resolved 55 bugs in the last week, fixing 30 of  these, including 5 new features and 1 PDT+.  For details, see our resolved bug list.

 

 

Lowlights

Dave Hyatt is once again feeling a lot of pain in his hands following his massive changes to CSS/XUL files for chrome reorg.  He'll have to stop typing for a while, and could use some help on code entry  from fellow seamonkeys.  This really kills our current schedule, since Dave is on the hook for 12 solid days of work between now and 5/4.  We'll have to come up with a new plan for finishing this work on Monday, 4/24.

Priorities
 
 

pinkerton:

11586  XP Menus need to be scrollable

saari:

Look into mailnews focus issues on linux (35006).

32344 scrollbar accepting focus.

36470 and 24525; problems on multiple monitor Macs.

hyatt:

Redistribute M16 features to the rest of the team.

Help everyone complete these tasks.

evaughan:

32155 [feature] Fix Stack z order

Check in new stylechange fixes and table fixes.

Implement spans for grids.

Help RodS with Gfx scrollframe problems.

pavlov:

Look at commercial UI drawing problems on unix (bug #33092)

Help hyatt on his features.

danm:

31867 [Feature] Don't hardcode navigator.xul for window.open()

Evaluate and check in helpful Robert O'Callahan's latest patches, which have been waiting for at least three weeks.  (He's been incredibly patient though)

Bug 35643, an avoidable but surprising crash.

Bug 19221, keep windows from popping up on top of

Macintosh modal windows, because this freezes the UI.

Tracking:

Durations of remaining XPToolkit feature work for PR2:

Last week:

DanM:  1 d
EVaughan: 6 d
Hyatt: 2.5+ d
Pavlov: 4 d
Pinkerton: 2 d
Saari: 1 d
Total: 16.5

Current: (individual assignments will change 4/24)

DanM:  1 d
EVaughan: 7 d  --> includes 5 d planned in M17 for PR2.
Hyatt: 12.5 d
Pavlov: 0 d
Pinkerton: 2 d
Saari: 0 d
Total: 22.5

Estimated feature work completion dates:  (details will change 4/24)

DanM:  4/26
EVaughan: 5/12   --> includes 6 d planned in M17 for PR2.
Hyatt: 5/4
Pavlov: done
Pinkerton: 4/28
Saari:  done

Issues

• Focus and key regressions are popping up again, mostly caused by "unrelated" changes. These are sucking up a lot of unplanned time, and hurting our schedule.  If you're making changes that could possibly affect these areas, please test thoroughly, and consider getting a review from Chris Saari.
• joki has taken ownership of the xp cursor bugs.  These bugs

are related to being able to set the cursor of a window (i.e. busy)
and have it keep links, etc from changing the cursor. (bug #9668)

Highlights:

• The team is focused on feature work for FCS (M16)
• Kin is back!!
• The Composer team met with folks from Intel. We discussed what it would take to provide the net with a minimal text widget editor. Simon did some= initial analysis on the current code base to see what we could eliminate from the top. Mike is working on the fundamental issues of providing text= widget editing through anonymous content.
• Akkana made some significant fixes for mail compose this week
• Charley attended the UI engineering conference this past week
• Joe is wrapping up block transformation issues, after this week the edito= r behavior should be pretty solid
• Kathy is focusing on getting drag-n-drop implemented
• 26 editor bugs were resolved this past week
• 104 open M16 bugs, 42 of those are feature work bugs

Lowlights:

• There are currently 283 open Composer bugs (M16-M30)

Priorities for M16:

• Kin:
• 6802: language selection interface for spellcheck
• 23682: wrapping for spellcheck
• Kathy:
• 33293: Composer prefs wording
• 23593: implement Composer prefs
• 6255: drag-n-drop
• Joe:
• 14753: list transformation (changing list type)
• 28456: removing bullets from a list deletes content
• 28716: multiple bullets on same line
• 32733: changing list style terminates list
• 26163: finalize typing rules
• 22331: block transformation
• 29843: nsHTMLEditRules::AfterEdit needs to ensure that the caret is alway= s made visible again
• IME support
• Simon:
• 34503: editor menu items should use command nodes
• 14662: implement context menus
• 24686: need to hook up copy in context menus
• 14768: disable animated gifs
• 25161: disable form elements
• 29162: disable frame elements
• 34