Summary
-
The XPToolkit team resolved 54 bugs in the last week, fixing 29 of
these. For details, see our
resolved
bug list.
-
Steady progress on features (e.g., trees running on box grid with few outstanding
bugs). Some slippage, but should still easily make the deadline.
-
Found a major performance bottleneck in Mac updating code.
Highlights
Mike Pinkerton (pinkerton)
Added support for dragging urls to the desktop on win32 to create internet
shortcuts.
Found a major slowdown in mac updating code that slowed down scrolling
and page loading/back/forward. basically a non-rectangular region was turned
into a bounding box rect and we repainted everything in the box.
Verified xpMenus, xbl, trees, and all xul widgets are up to snuff with
dom-2 namespace changes.
Added cmd-click to open a link in a new window
Chris Saari (saari)
Worked with mjudge to fix some focus issues with Ender Lite.
Daniel Matejka (danm):
Fixed some resizing bugs.
Eric Vaughan (evaughan):
Fixed numerous ender light problems.
Fixed grid bugs to support tree work.
Stuart Parmenter (pavlov):
Fixed some window title bugs, several event queue bugs, and assorted other
defects.
David Hyatt (hyatt):
Implemented lazy scrolling in the new tree widget.
Implemented keyboard navigation and ensure visible in the new tree .
Brian Ryner (bryner)
-
Fixed several file picker bugs, and some tree selection problems.
-
Got classic skin to install all files on unix (currently just using the
windows files).
Lowlights
-
Landing of the tree rewrite didn't make 6/9, rescheduled for 6/14.
The bulk of the work is substantially complete, with handling of exception
cases and further testing remaining (details below).
Priorities
pinkerton:
39699 No
submenus if scrollbar
28226 Auto-scroll
tree during drags. Prep work for drag scrolling in trees (get mac reflowing
during a drag)
saari:
39633
Quit
from browser or composer doesn't prompt to save Compose
38673
Insertion of certain items as first element disables buttons
hyatt:
41275
BLOCKER ENDER-LITE, typing in subject line of mail causes entire page
41878
setAttribute for checkbox not working
30511 Convert
<tree> to <grid>
32156 Grid
widget lazy scrolling
Continue to work on the tree widget, getting it ready to land
hopefully by Wednesday.
evaughan:
32730
Auto-scrolling XPMenus
Tree and Ender issues.
pavlov:
account wizard not appearing when 3 pane is launched bug 40737.
danm:
25684
-- networking alert dialogs need a real window for their parent, not HiddenWindow.
35726
-- untangle our appropriation of some extra identifiers in the JS window
namespace from the scripts that are already using those names for their
own misguided purposes.
garyf
41136
Crash while closing aim after signoff
bryner
29027
Minimized not restored when called from Taskbar or Tasks menu.
22794
Wheel mouse is not working/Genius NetMouse
Dogfood:
39392
hyatt ETA: 6/14 Crashing when adding buddies to buddy list
39699
pinkerton@netscape.com 6/16 No submenus if scrollbar
Beta Stoppers:
61 nsbeta2+ bugs, but only 26 are without a cutoff date, so we should
be in better shape on 6/16, much better on 6/23.
Risk:
The tree rewrite has already slipped, since it was due on 6/9,
but it should land by 6/14. Current open issues in the new tree are:
-
Scrolling asserts and crashes on occasion.
-
Mailnews crashes when deleting a message.
-
Message compose textfields don't work in new tree (this is with the old
Ender. Hopefully Ender Lite will work. Will update as we find
out).
-
Columns cannot be resized. (Evaughan has fix but is on vacation.)
-
Columns cannot yet be collapsed. (Doesn't work in current tree either,
so we don't view this as blocking landing.)
-
Column sizes incorrect. (Evaughan has fix but is on vacation.)
Barring unexpected dogfood+ bugs, other features should land by 6/16.
Features:
ID |
Owner |
Result |
StatusSummary |
Summary |
32730 |
evaughan@netscape.com |
|
[nsbeta2+]Exception Feature 5 days NEED ETA
<MM/DD> |
[feature] Auto-scrolling XPMenus |
30511 |
hyatt@netscape.com |
|
[nsbeta2+] Estimate landing on 6/14 |
[feature] [XUL] Convert <tree> to <grid> |
32156 |
hyatt@netscape.com |
|
[nsbeta2+] 4 days NEED ETA <MM/DD> |
[feature] Grid widget lazy scrolling |
28226 |
pinkerton@netscape.com |
|
[nsbeta2+] 4 days NEED ETA <MM/DD> |
[feature][tree d&d] Auto-scroll tree during a drag |
Highlights:
-
The team continues to focus on M16 dogfood+, nsbeta2+, and regression bugs
-
The team also completed their self assessments
-
Anthony jumped in with both feet into memory leak issues
-
Akkana resolved the multiple output with table cells problem - Yippee
-
Kin discovered the problem with the invisible text insertion
-
Simon is stuck in crash nightmare mode, been working through multiple Mac
crash issues
-
Mike checked in ender-lite
-
Joe fixed "pain-in-the-@$$" backspace deletion problem
-
Kathy checked in drag and drop for Composer -- Ya Hoo!
-
Charley is plowing through dialog box issues
-
43
editor bugs were resolved this past week, of those 9 were marked nsbeta2+
-
18
open M16 bugs -- we are actually under 20 m16 bugs!!
Lowlights:
-
There are currently 277
open Composer bugs
Dogfood:
41206: Invisible text on insertion. There is still a bug in the interaction
with the ViewManager and Linux implementation of nsWindow. Kin is working
on this now.
41339: File/Quit crash on Netscape6-mac.sea.bin. Simon believes it is
a memory allocation issue, awaiting feedback on reproducing the error
42210: Dialog boxes are opening the width of the screen and the Ok/Cancel
buttons are not accessible, this is normally a XUL box issue, being debugged
now.
Beta Stoppers:
bugs mentioned above and
41744: Composer creates non-compliant HTML -- with all of our chest
thumping about standards compliance, this would be a very bad thing to
let happen.
Risk:
no major risk issues at the moment.
Features:
39308 & 14874: Context menus, resolving these two open issues are the
priority for Kathy this week
Tables: 1 open bug (20973), Charley needs to wrap up the dialog box.
Priorities for M16:
-
Kin:
-
35899 -- Page scrolls on selecting text in position:relative div
-
41206 -- text inserted is initially invisible
-
41406 -- copy text to empty fields shows invisible
-
Kathy:
-
14874 -- Each edit control needs a context menu
-
39308 -- Context menus options are not functional
-
Joe:
-
41744 -- editor can create non-compliant HTML
-
41746 -- paragraph format feedback incorrect
-
Simon:
-
41339 -- file/quit crash on mac
-
41681 -- crash quickly closing new editor window
-
42145 -- crash on sonydirect.com
-
Mike:
-
19392 -- blur doesn't work
-
34896 -- text widgets should use anonymous content
-
39528 -- nsDOMSelection::ToString with OutputSelectionOnly is busted
-
41206 -- text inserted into Composer is initially invisible
-
Charley:
-
20973 -- implement remainder of table editing -- dialog box work
-
38875 -- disable toolbar and menu items while in HTML Source edit mode
-
39919 -- DOM_L2: Does the editor rely on old incorrect DOM Level 1 behavior
-
40204 -- Composer color picker swatches on toolbar don't update correctly
-
41735 -- Checkmarks in Composer menus are missing
-
42210 -- dialogs render unusable
-
Akkana:
-
16311 -- line selection
-
32807 -- paragraph selection
41916 -- copy/paste within a table sometimes picks up container
Issues:
-
Finding lots of bugs that are exposed within Composer, but end up
being the result of more low-level issues. The frustrating part is that
the perception is that Composer is not usable.
People:
-
working on getting an intern, hopefully
Accomplishments:
-
Kin
-
Did some poking around the editor code for jst, to verify that his ImageElement
changes would not break the editor.
-
Code reviewed brade's composer D&D changes.
-
Checked in fix for bug #41605 and #41734. This was due to a missing AddRef
in GetSelectedElement().
-
Wrote simple test cases for several bugs to make it easier to debug and
test.
-
Tried to help mscott figure out why IMAP mail crashed when loading my inbox.
-
Checked in changes from BlakeR1234@aol.com for bug #41283.
-
Started rewrite of nsSelection::DoAutoScroll() to fix bug #35899, but got
sidetracked several times by dogfood bugs that popped up.
-
Spent some time Friday morning writing down self-assesment accomplishments
at beppe's request.
-
Triaged and did preliminary debugging for various bugs.
-
Currently trying to fix Linux redraw problems that occur in Composer and
TextFields. (Bug #41206, #41406, #41840)
-
Simon
-
Spent lots of time looking at memory leaks, trying to determine the cause
of bug 41681 (crash closing editor window quickly). Found the following
leaks (41695 et al.)
-
nsParser leaked via the mParserFilter (also found by Rusty in bug 40149)
-
Mac menu code leaked like a sieve; I rewrote parts of it to use weak references,
and fix some obvious refcounting deficiencies (still needs to be checked
in).
-
Composer still suffers serious leaks (bug 42098) -- a leaking nsXULDocument
is causing all XUL document content, including the nsEditorShell and nsEditor,
to be leaked.
-
Modified and checked in some code from an external contributor to implement
support for the URL Echo Apple Event (bug 36840).
-
Found a nasty memory trashing bug (41996) that could have caused lots of
random crashes.
-
Worked with ftang on an IME bug, caused by a focus-related problem (41573).
-
Reviewed some code by Gary Wade for Mac font issues
-
Found a fix for an MSL bug that crashes JPEG code (41360)
-
Pestered waterson to fix some string performance issues in CSS code (41321).
-
Mike
-
turned ender-lite on
-
worked with hyatt and brendan on trying to resolve the address line problem
in mail, is now in the hands of the mail team
-
resolved 10 or 12 bugs from the m17 list
-
fixed another dogfood bug 41814
-
fixed several nsbeta2+ bugs (19392, 34896, 39528,
-
Anthony
-
Joined team on the 6th.
-
Moved cube, set up new machines.
-
Pulled new Seamonkey(Win and Mac), and Nova trees(Win only).
-
Fixed my warnings on Linux (waiting on my Linux tree before checking in
last batch of fixes).
-
Installed Quantify.
-
Kathy
-
Checked in drag and drop within an editor window (thanks to Kin for reviewing/testing)
-
Helped Simon out testing his fixes and sending him my fixed for some of
the memory leaks in editor and mac-specific code
-
Reviewed editor bug list for candidates to be pushed off to Future (or
those which should be resolved before RTM)
-
Self Assessment
-
Joe
-
dogfood+ deletion bug: done and checked in to tip and M16.
-
deletion bug with no #: you couldn't easily delete empty lines that were
right next to block elements (like lists or tables). fixed.
-
working on 41744, 41746, 40363, 40364.
-
also working on bug with no #: <hr> is being treated like an inline,
but it's a block. I need to split paragraphs, etc., when putting in an
<hr>.
-
another one with no #: you lose fonts and other inline styles when you
start a new block. i.e., make a paragraph, choose bold, type some text;
now hit two returns to start a new paragraph. keep typing: you are no longer
bold.
-
Charley
-
List submenu (41034),
Object properties ( 41351),
Table selection optimization (41045)
-
Fixed nsbeta2 bugs: Checkmarks for menu items ( 41735
), Duplicate DTD strings (41511),
Added a pref for autocopy when selecting with appropriate platform defaults
(41045).
-
Made great progress on Table Properties dialog (20973)
XUL is done, including new checkboxes to control which attributes are applied.
Most of the JS (rest described below).
-
Akkana
-
Fixed bug 40457, the multiple output when table cells were selected:
-
Added a new Range API (nsIDOMNSRange) to expose some of Joe's utilities
for determining when a node or point is in a range
-
Helped test and debug Ender Lite, and flipped the switch on Unix.
-
Verified that three of my bugs that I suspected would be fixed with ender
lite are indeed fixed.
-
Fixed 41916 (paste of something within a table cell picks up the table
cell): currently awaiting review by Rickg.
-
Discussed better longterm solutions with Rick, and filed 41868 as a reminder
to do this after beta2 is over, dependent on the fix Rick is providing
for 41825.
-
Ate dogfood, filed bugs, commented on other bugs, helped track down morning
blockers.
-
Helped with a few Linux installs/setups.
-
Triaged my m17 bug list.
-
Self assessment.
Summary
-
Heikki Toivonen got his H-1 visa approved also! Both, Heikki
and Johnny will tentatively start at Netscape on July 5.
-
Tom Pixley will be on the Netscape campus from June 22nd through 29th.
-
The team is cranking away on nsbeta2+ bugs.
-
Harish is the only team member without any nsbeta2+ bugs. He has
volunteered to help Johnny out. Thanks, Harish!
Highlights
Harish Dhurvasula (harishd)
Checked in fixes to 7 nsbeta2+ bugs!
Fixed a MLK, in nsTokens, caused on international sites.
Moved 40% of my bugs to future.
Helped Rickg with StrictDTD.
Working on a regression test for NavDTD.
Helped Bindu with regression tests.
Tom Pixley (joki)
Is testing fixes to 4 nsbeta2+ bugs (24698, 25117, 38177) in his local
tree. Will check them in early this week.
Has a partial fix to nsbeta2+ bug 29517.
Plans to be done with his existing nsbeta2+ bug this week.
Johnny Stenback (jst):
-
Fixed nsbeta2+ bug 30249, made mozilla compatible with 4.x wrt the type
of varios image element properties.
-
Fixed nsbeta2+ bug 40514, bad parsing of the options argument passed to
window.open(), caused mozilla to ignore some options.
-
Fixed nsbeta2+ bug 41718, case sensitivity problems in the docshell with
javascript: URL's.
-
Fixed nsbeta2+ bugs 17351 and 33018, problems with w.document.write() after
w=window.open().
-
Fixed nsbeta2+ bug 41487, problems with form controls with the same NAME
and ID.
Nisheeth Ranjan (nisheeth):
Away to the XML Query face to face meeting on Monday and Tuesday.
Met with Radha and Rick Potts about session history interaction with docshell.
We're all on board with the pseudocode and have divided the implementation
between us. Radha will handle the session history side of things.
Rick will do most of the work on the docshell. I will make a few
changes to the docshell so that page state is saved to and restored from
session history.
Interviewed Edward Kandrot
Finalized re-imbursement for Johnny and mailed an invoice to AOL accounts
payable.
Triaged Clayton's bug list from 34 to 24. More work on that this
week.
Did a pass through my bug list and marked bugs "future".
Lowlights
None.
Dogfood:
No dogfood bugs.
Beta Stoppers:
28 nsbeta2+ bugs. Everybody has a good handle on their nsbeta2+
bugs. We had a meeting on Friday where we did some bug distribution
to balance the bug load on each team member. This should show up
this week once we go in and re-assign bugs between ourselves.
Features:
-
Dynamically add linked stylesheets (jst, bug 7515): Johnny plans on working
on this next week. Hasn't done any work on this yet.
-
Support JS LiveConnect calls in existing content via upgraded plug-in binaries
(jst, bug 38495): Johnny is waiting for the code that lets us do calls
on the plugin through XPConnect (bug 32150 assigned to Mike McCabe).
Issues:
None.
Escalations:
None.
People:
Heikki Toivonen got his H-1 visa approved also! Both, Heikki
and Johnny will tentatively start at Netscape on July 5.
Tom Pixley will be on the Netscape campus from June 22nd through 29th.
Summary
-
Highlight-
-
Several bug fixes.
-
impressive net contribution: mng decoder from Tim Rowley (tor@cs.brown.edu)
-
We did our first ever Necko/Imglib outing!
-
Lowlight-
-
A little high on nsbeta2+ bugs. Need to clean these up soon.
Detailed Status
Highlights
-
gordon
-
Took part in the first ever Necko/Imaging Offsite 2000 Pro.
-
Worked a bit with Patrick on PSM for the Mac...but not as much as he would
have liked. I should have more time next week.
-
ruslan
-
Not much to report because of JavaOne.
-
Partially fixed: 41248, 41876 - we had crashers in the networking induced
by bugs in the upper layer. However we're handling these error cases correctly
now. Also implemented our portion of SSL proxy connect logic (40203). Now
somebody has to continue on the PSM side.
-
neeti
-
Fixed bug 41764
-
Investigated/resolved 41253, 41172, 39116,41624, 40235, 2694, 39957
-
pnunn
-
bug#35316: found a bug in the mac nsStdlib bug.
line 89, file_io.mac.c, last next in linked list is garbage.
Opened bug#41360: assigned to sfraser, cc:scc.
-
bug#32394: closed/invalid. corrupted gif stops animating.
-
bug#25450: closed/duplicate of #30267.
-
bug#35946: closed/dupe of #35316.
-
bug#41316: closed/invalid. Reopened as bugscape #1133
-
We are getting a mng decoder from Tim Rowley (tor@cs.brown.edu)! Yea tor!
-
bug#41634: created simple test page. and studied problem.
-
ART: testing new mac art library changes from retsbol@aol.com
-
andreas
-
Not much to report: Went through my bug watch list, commented on some,
confirmed a few other bugs. Other work was indirectly necko related (walllet,
bug 41702) or non necko related (xslt).
-
gagan
-
attended JavaOne on later half Wednesday.
-
organized the group outing (phew!)
-
juggled with bugs.
Lowlights
-
andreas
-
Bug 19313 is blocked by bug 31225
-
gordon
-
Spent a few days trying to reassemble my Linux box after switch with David's
old machine. It's almost working again.
-
I was unable to make progress on Async DNS for Unix because of the above
proglem.
-
pnunn
-
gagan
-
Haven't worked on any code this week... :-(
Dogfood Bugs
-
gagan
-
42073 (Cannot load XUL files using file URL) This seems like a regression,
nothing specifically has changed in the necko part of file handling. Gagan
is investigating this. Most likely this is a mstoltz bug.
-
42076 (Non-chrome URLs not able to load chrome URLs) possible dup of the
one above.
Beta Stoppers
-
31
bugs. 11+ Several have partial fixes. The current list is under control.
Risks
-
Just a little high on nsbeta2+
-
The SSL Connect problem is fixed from necko side. We need to now find an
owner in the security area to own this for the remaining part. I (Gagan)
will chase down the new security area owner!
Features
-
AsyncDNS (bug 10733)
-
Proxy Auto Config (bug 20145)
Issues/Escalations
-
andreas
-
Need general direction on where to go with bug 32895
-
pnunn
-
put in vacation time request for last week of June
- Basic support for the default namespace. This is not completely finished.
- We're using the right namespace when creating DOM elements in Mozilla.
This allows Mozilla to choose the appropriate type of element (eg HTML,
MathML, ...) and thus display the output correctly.
- Eliminated DOMString.
- Andreas Otte found the bug in nsXMLContentSink that prevented a processing
instruction to get Mozilla to call Transformiix (see bug
http://bugzilla.mozilla.org/show_bug.cgi?id=36038). Thanks Andreas!
This all means we now are very close to having a working Transformiix module
in Mozilla. The Transformiix module already transforms simple testcases and
the result displays in Mozilla.
Some issues have come up that i'd like to discuss/clear up. Right now,
Mozilla creates a document and passes that to Transformiix. It also passes
two DOM trees that were created in the context of that document, but aren't
hanging directly from the document. Mozilla expects Transformiix to hang the
output tree from that document. Here are the issues we have noted, and some
of my comments.
- Transformiix can't set the type of the result document
Andreas wrote: "We know nothing of the mimetype of the resulting document.
Can be text/xml, can be text/html, whatever. I think we need to create the
right document and the right ContentSink to layout the resulting document
correctly."
I don't think this is a problem. We get a XML document from Mozilla and
Transformiix constructs a result DOM tree within that document. No problem
for XML output. Transformiix uses a default namespace of HTML if the output
method is HTML, and Mozilla correctly creates HTML elements. These can be
added in an XML document, so i don't think there's a problem for HTML
either. I haven't looked closely at plain text output, put i guess we could
just create Text nodes in the XML document?
- Transformiix doesn't get source/style documents from Mozilla
The XSLT specification states that an XSLT processor should transform a
source tree into a result tree. So Mozilla might be right in passing us
trees instead of documents. However, at the root of that tree there is no
document element. This causes all sorts of problems. The most blatant is
that the XPath location path "/" doesn't work. XPath specifies: "/ selects
the document root (which is always the parent of the document element)".
There is no document root, because the start node of the tree we get isn't a
document element, so we can't get "the parent of the document element".
Another problem is that we can't get the URI of the stylesheet document
(needed for xsl:include). If we get the document of an element in the
stylesheet tree, we get the original XML document, and thus the URI of the
XML document. There is no way to get the URI of the stylesheet document.
Previous Updates
|