status update

maintained by Tim Rowley <tor@cs.brown.edu>

Last Updated Friday August 18, 2000

This status update page is updated every weekend. To get updates and news throughout the week you are invited to check out mozillaZine, a site devoted to Mozilla advocacy.

Previous Updates


Friends of the Tree

  • Tomi Leppikangas for hacking on linux performance - Stuart Parmenter
Module Updates
MailNews Performance
August 11
Submitted by Pratik Dhupia <pratikd@netscape.com>

MailNews Performance test results for Win32, Red Hat Linux 6.1 and MAC OS have been updated and can be viewed at http://www.mozilla.org/mailnews/ under Quality Assurance.

Tested on comm build 2000-08-09-08, M18.

MailNews
August 14
Submitted by Steve Elmer <selmer@netscape.com>

Issues

  • Very high nsbeta3 incoming rate
  • Current M18 builds are very unstable all-around

Vacations & Sabbaticals!

  • JeffT 7/10-9/5
  • SSpitzer 7/5-8/26
  • RHP 8/18-8/27

MailNews

Profile Manager

  • now have triaged list of nsbeta3+ bugs for profile manager/activation

CCK

  • small amount of bug work going on
XML/DOM
August 14
Submitted by Nisheeth Ranjan <nisheeth@netscape.com>

Summary

  • The initial triage of our beta 3 nominated and untargeted bug list is complete.
  • Our team has 74 nsbeta3+ bugs.

People

  • Peter Van der Beken from Belgium is on the Netscape campus for two weeks starting today.  He will be hacking on XSL.  Come down to the Gecko area to say hi!
  • Tom is on sabbatical till August 25th.

Accomplishments

  • Harish Dhurvasula (harishd)
    • Triaged my bug list & rickg's bugs ( 16 bugs )
    • Checked in fixes for 13 nsbeta3+ bugs (  12 more left )
    • Ran trace-malloc for arena based token cache ( analysis still pending ).
  • Heikki Toivonen (heikki)
    • Resolved 6 bugs: 38647 (fixed), 29473 (fixed), 41772 (fixed), 47930 (fixed), 46536 (invalid), 46597 (worksforme). Was able to finally check in the last bugs with fixes in them as well as fixing one new.
    • Triaged joki's untargeted bugs with saari & ckritzer. Me and saari took the nsbeta3 nominees on our plates so joki does not have any nsbeta3 bugs on his list. (There are 9 bugs on joki's nsbeta3+ list that have been fixed, saari has not yet marked them fixed).
  • Johnny Stenback (jst)
    • Bug triage, bug triage, bug triage and then some bug triage.
    • Finally turned on the new DOM Level 2 changes, yay ! ;-)
    • Worked on DOM API string changes, this looks pretty good right now, the code compiles and mozilla runs without any obvious changes.
  • Nisheeth Ranjan (nisheeth)
    • Triaged beta 3 nominated and untargeted bugs.
    • Attended a day long class titled "Communicating for Success" on Thursday.
    • Attended the client platform meeting organized by Vidur.

Plans for next week

  • Harish Dhurvasula (harishd)
    • Analyze trace-malloc data for token cache
    • Fix atleast 5 beta3 bugs ( enabling NOSCRIPT being the top priority )
    • Bug triage ( includes Clayton's list )
  • Heikki Toivonen (heikki)
    • Fix nsbeta3+ bugs.
  • Johnny Stenback (jst)
    • Land the DOM API string changes if I and Vidur have time to make the remaining changes (minor changes compared to what we've done so far)
    • Work on my nsbeta3+ bugs !!!
  • Nisheeth Ranjan (nisheeth)
    • Finish focal reviews.
    • Prepare and give talk on "Gecko and Open Source Development" to the summer interns.
    • Start fixing nsbeta3+ bugs.

Lowlights

  • None.

Escalations

  • None.
XPToolkit
August 14
Submitted by Peter Trudelle <trudelle@netscape.com>

Summary

  • The XPToolkit team resolved 63 bugs in the last week, fixing 34 of  these, including  25  '+'.  For details, see our resolved bug list.

Highlights

  • Mike Pinkerton (pinkertoni)
    • Rewrote tree drag scrolling to not use timers or view capturing, simplifying it and avoiding crashes (43428)
    • Took a patch from dean_tessman@hotmail.com for menus not working properly with several levels of submenus open
  • Chris Saari (saari)
    • Landed Joki's changes and resultant leaks, although dbaron says I didn't get the  secondary smaller cycle yet... of course I don't see how there could be  a cycle there, but I'm working on it.
    • 28508 Select All doesn't work from context menus
    • 39308 clipboard items in context menus don't function (focus loss)
    • 41325 shortcuts for back/fwd don't work on winNT
    • 43309 Tab & Shift+Tab takes 5 extra tabs to get to the check box i
    • 44257 -Pressing tab in url bar causes cursor to dissapear
    • 44503 Tab order not obeyed
    • 48126 60k of leaks added to smoketest
  • Daniel Matejka (danm)
    • 28193 window.open() ignores 'scrollbars=no' parameter
    • 31516  Unable to create offscreen window OR unmapped XUL window
    • 47491 Crash closing window
    • 48066 closed windows are not closed completely, just hidden
  • Eric Vaughan (evaughan)
    • On vacation
  • Stuart Parmenter (pavlov)
    • 46729 - Have a fix to make smart quotes paste as plain text normal quotes when they need to. I will check in as soon as nhotta checks in fix for 48284
    • 26502 - linux performance.
      • remove pointless call to XSetFont every time we drew text
      • make images hold on to a GC so they don't XCreateGC XCopyGC XCopyArea XFreeGC every time they image is drawn. (not checked in yet)
      • avoid pulling an event out of the event queue and immediatly putting it back in when typing.
    • 43580 - removed a few more static nsCOMPtrs in nsClipboard code
    • 47921 (nsbeta2+) - fixed linux installer on machines where /sbin/sh is bash2.
  • David Hyatt  (hyatt)
    • Enabled xbl to load asynchronously
    • Enabled xbl to include css (fixed 21890)
    • Taught xbl about scopes (who the originating binding is, needed for 46505)
  • Dan Rosen (dr)
    • 38787 - fixed, but hyatt notes that the chrome cache doesn't canonify urls, and therefore thinks two equivalent urls are not the same; performance associated with canonifying for each cache access is a new bug, 48104.
    • 42635 [nsbeta3+] - fixed: blew away c++ replaced by xul/js and lost my tree virginity in the process by breaking win32 builds.
    • 46658 - simple fix, but i was confused for a while not knowing that win32 and mac use native filepickers. opened new, related but 48311 for an xp filepicker problem.
  • Brian Ryner (bryner)
    • Fixed mousewheel bug 42294 (font resize affected chrome) Dhecked in contributor patch from Jan Varga for filepicker navigation and i18n
    • Fixed 48752 (filepicker regression from nsDirectoryService changes)

Lowlights

Priorities

  • pinkerton
  • saari
    • Mac performance ASAP
  • hyatt
    • Tree bugs
    • 41138 Dragging address to compose window crash after 3rd address
    • 45248 profile mgr crashes if you use <ret> key to pick profile
    • 45728 crash if maneuvring with keys in preferences / fonts
  • evaughan
    • more vacation
  • pavlov
    • 26502 - linux performance
    • 33313 - css2/3 system fonts on unix.
  • danm
    • Measure the leak in bug 48066 so it can be triaged.
    • 28663 choose app window does not close on exit
    • 30394 Window.open() parameter ignored
  • dr
    • 47395 M18 [nsbeta3+] perf: nsXULKeyListenerImpl::HandleEventUsingKeyset accounts for 20-25% of typing time (this is my "interesting" bug, given to me mercifully by hyatt, but i will need a day or two to learn what this code is doing in order to fix it).
  • bryner
    • 42109 crash at shift-clicking a title in the messages list
    • 47795 Linux filepicker is slow to load large directories
    • 48311 Filepicker does not display proper file list

Dogfood

  • None

Beta Stoppers

Risk

Issues

Escalation

People

Architecture
August 14
Submitted by Warren Harris <warren@netscape.com>

Issues

Engineering Status

  • waterson
    • Helped dbaron wrap up 45676  (scrollbar hover leak). This was a hard, hard leak, and dbaron deserves most of the credit.
    • Worked up fixes for the following bugs:
      • 44480 div not resizing to window; we were using width of <html> element, not <body>
      • 47154 entities cause whitespace to disappear; turns out that \n and \r not treated as whitespace
      • 48086 lose selection after sort; regression caused by fix to 46616
    • Conference call with IBM bidirection text layout group. Their changes went onto a branch this week to prepare for "touch down" (i.e., under #ifdef) early next week.
  • vidur
    • Spent most of the week working with Johnny Stenback on DOM string  changes. We're changing the nsString&'s in the DOM API to nsAReadableString& and nsAWritableString& to get rid of problems with fragility and allow us to use different string implementations where appropriate. The changes touch a large number of files and should land sometime this week.
    • API Review for the embedding interfaces. We're done with our first round of reviewing! Judson's group is going to make the IDL changes and send them out for first review very soon.
    • Footprint/Performance management meeting.
    • Held a meeting on remoting XPCOM. Jband is looking into this as part of a web application story.
  • rayw
    • Lots of communications with developers requiring or affected by XPCOM changes WRT nsbeta3, resulting in a more-complete picture.
    • Better triaging for XPCOM.
    • More progress on the changes to APIs and progids.
    • Posted text regarding progid -> capability changes.
    • Expect early next week to recieve finished target APIs and progids, which I have worked on up until now.
    • Another issue was the plugin and string bundle registrations in the component registry.  I feel that perhaps we can just use a new registry and leave these things to others to clean up -- they can actually remove the oldregistry when it is clean.  We don't have to push them into it.
  • waldemar
    • Worked on JavaScript 2.0 language definition and on tools to allow me to maintain separate specs for ECMAScript Edition 4 and JavaScript 2.0 while using a common document base.
    • Investigated several Mozilla CSS1 bugs.
    • Arranged short ECMAScript Edition 4 meeting with Microsoft on August 22nd.
  • rpotts
    • should return from vacation next week
  • jband
    • Trying to focus on footprint and performance bugs. So far these include: 48855, 46707, 46703, 43902. I expect that more digging will reveal other good places where I can help make improvements.
    • Also did some work on JavaScript and XPConnect bugs: 48564, 47354, 39808.
  • warren
    • Landed changes for hashtable string keys. Saved 1M of space to startup and load the handful of tinderbox test pages.
    • Still trying to get jar packaging up and running. Unfortunately, things have bit-rotted considerably since I had it working on Windows and Linux last week -- trying to debug what's going wrong now.
  • dprice
    • on vacation until 8/14
  • scc
    • made a lot of progress on multi-fragment strings
    • fixed up BeginReading, et al, routines to be efficient for multi-fragment strings
    • wrote the underlying chunk-list management for multi-fragment strings; more to follow
    • got multi-fragment strings limping along; they pass all tests except those related to Substring and nsPromiseConcatenation
    • figured out how to make substrings and concatenations work for everybody (and improvements to iterator), but haven't implemented yet
    • supported vidur and jst's effort to convert the DOM APIs
      • rewrote NS_NewAtom to work with readables, though I'm not sure how vidur and jst want to get these changes
      • bug #48464: fixed nsAWritable[C]String assignment problems
      • wrote a character sink based on nsCRT::HashValue to support adapting NS_NewAtom for nsAReadableStrings as part of Vidurs DOM conversion work
    • fixed bugs
    • reviewed, tested, and/or checked-in other peoples patches
    • wrote documentation
      • added more examples of using iterators to modify strings in place, and thoughts on copy_string drawn from an email discussion
      • wrote and emailed the comparison of Opera and Mozilla, which I needed to do for mozilla.org
      • attended meetings
      • mozilla.org staff meeting (1 hour)
      • API review meeting (1.5 hours)
      • making us not suck on the Mac (with PDT) (1.5+ hours)
  • kandrot
    • no status
Necko/Imagelib
August 15
Submitted by Gagan Saksena <gagan@netscape.com>

Summary

  • RJC joins Necko/Imglib to help with FTP!
  • nsbeta3+ still high and there are still nominations coming along.

Dogfood and Risks

  • None

Beta Stoppers

  • 29 across a team of 6 developers.

Highlights

  • ruslan
    • Fixed 47408 (add versioning to the cache), 45337 (nsHTTPChannel looses track of security info),
    • 46496 (Header value parsing ("chunked") isn't HTTP compliant: case-sensitivity)
    • 42459 (worksforme as the cause of the leak was long fixed as part of the other bug)
  • rjc
    • 47927: investigating
    • 35022 & 35817: in-progress work on RDF performance optimizations
    • 44324:  display error msg if overloaded FTP server
    • 31586:  URL resolution when logging into user's home directory via FTP
    • 26131:  sorting performance optimization
  • neeti
    • Closed bug 30670
    • Investigating Cache Correctness bug 47400
    • Working on bug 21879.
  • pnunn
    • Bugs Closed: #47411, #74793, #41040
  • gordon
    • Investigated untargeted bugs (40076, 26718, 30670)
    • Investigating potential cache modifications (47405)
    • I've noticed some differences between 4.x and Mozilla in what requests made on reloads.  I'm not sure what the correct behaviour is supposed to be.
  • gagan
    • Updated status/bugs with correct queries.
    • Triaging bugs for the whole team.
    • Working on about:cache implementation.

Lowlights

  • pnunn
    • sick from 08-08 to 08-11.

Issues/Escalations

  • ruslan
    • 48358 (www.hotmail.com is blank) - it turns so that the redirect coming from the new IIS5 is grossly http-incompliant and doesn't send the newline past headers wich confuses mozilla
LDAP
August 18
Submitted by Dan Mosedale <dmose@netscape.com>

Unix

  • Added a debugging frob allowing users to specify various NS_ASSERTION behavior by an environment variable, including dumping a stack trace to stderr or suspending the process so a debugger can be attached. See http://www.mozilla.org/unix/debugging-faq.html#assertion for details on how it works.

LDAP

  • With help from the new debugging frob and Doug Turner, tracked down threadsafety assertion when callbacks happen on the LDAP connection thread.
  • Major fixes to the code that cleans up at the end of an LDAP search and/or when a search has been cancelled. Most visible effect: the throbber should always stop spinning at the end of a search, rather than sometimes lingering on indefinitely. Thanks to Judson Valeski for helping me sort this stuff out.
  • Helped Mike Shaver test fixes to the Javascript component loader so that the capability permissions on function callbacks are lexically closed, not just variable scopes. This unblocks the RDF datasource, so I'm now back to work on that.

Previous Updates