Heads up on priority items
- nsbeta1+: 42 (+4); nsbeta1: 31 (+6). We are going in the
wrong direction :( We need to resolve more than half of our current nsbeta1+
bugs by this time next week to stay on track.
- P3P: provides simple, automated
ways for people to control the use of personal information on web sites they
- harishd checked in P3P compact policy support, not on by
- nisheeth found and fixed ~400k leak in reloading a page
with lots of param elements.
- mstoltz gave security techtalk and wrote notes from the
RSA conference; usefuls reads to everyone.
- Checked in, on the trunk, support for enabling P3P on all three platforms
- Worked on correcting JS line numbers for HTML documents
[bug 111576]. The fix started out to be trivial but after uncovering a few serious
defects the fix turned out to be complex.
- Worked on P3P policy viewer (bug 124503). Spent sometime learing XSL.
- Did code reviews and attended meetings.
- Caught up with my email n' all after getting back from the W3C Plenary
meeting in France last week.
- Worked on iframe loading bug 52334, making good progress, still some issues to resolve.
- Did a bunch of sr's.
- Fixed bug
(XMLSerializer is broken)
- Fixed some fallout from darin's URI landing
- Worked on bug 34849
(dynamically added STYLE element doesn't alter style)
- Started reviewing XPath patch by Pike
- Did some other reviews
- Continued work on fixing bug 81446
. Fixed a leak of PARAM elements which now get destroyed when the document
goes away. This fix does away with 394206 bytes of leaks described in bug 129423, bug 129424, bug 129425, and bug 129429. We still leak other memory across reloads of the same page so bug 81446
is still open. More investigation this week.
- Fixed bug 125466
(Add AString, CString, UTF8String support to the XPCOM proxy event mechanism).
With this bug fix, I have no more string work left on my plate. Finally!
- Read paper on Meta Compilation (MC) titled "Checking
System Rules using System-Specific Programmer-Written Compiler Extensions"
(see http://www.stanford.edu/~engler/mc-osdi.ps). The paper talked about
using MC to write compiler extensions that automatically checked their code
for rule violations. The MC approach was used on Linux, OpenBSD, and two
other real world systems and found roughly 500 errors which led to numerous
kernel patches. It might be interesting to apply this idea to the Mozilla
codebase and check for leak errors and other violations of "higher level"
rules that incorporate Mozilla-codebase-specific semantics. I've written
to one of the authors of the paper, Dawson Engler, and requested him to give
us access to the MC tools developed by him.
- Was virtual Clayton for last week because Clayton was
out of town.
- Worked out bug list and priorities for Moz1.0
- Published notes from RSA conference
- Updated project plans on Component Security homepage
- Did Monday night tech talk on Mozilla Security
- Patch proposed to fix #119335.
- Worked on #125732 and uploaded a modified patch.
- Working on #125037.
- Spent some time for #107927 and #69792.
- Code reviews for harishd, peterv, rayw, firstname.lastname@example.org
- Checked in fix for 123953, crash when XMLSerializer
tries to serialize 'orphaned' node.
- Checked in fix for 129109, XMLHttpRequest.abort() was
only partially implemented.
- Landing page updates regarding P3P and iframe loading.
- Discussed the remaining string bug issues with jag
that were found while fixing bug 110544, 2.5% improvement in page load by improving comment parsing. We seem to
agree there is a bug, but we have a workaround so the perf fix can land.
- Made fix for 129608, parsererror should be in its own
- Bug triage.
- Starting to give out focal review comments.
Plans for next week
- Work on bug 124503
- Work on nsbeta+ bugs.
- Triage bugs.
- Finish up the iframe loading work.
- Finish bug
(dynamically added STYLE element doesn't alter style)
- Work on bug 94036
(Fix xsl:number support)
- Continue reviewing Pike's XPath patch
- Continue working on bug 81446.
- Profile slow DHTML test cases to see whats up.
- Sync up with tor to continue with image optimization
- Start working on Moz1.0 bugs
- Get the reviewers' guide online and begin assigning
bugs based on it
- o Fixing 125037 and other bugs.
- Finish with focals.
- Make sure mozilla1.0 and nsbeta1 numbers reflect reality
for the whole team.
- Check in the fixes I have in my tree.
- fixed bug 124042 "support internationalized URIs" (on trunk and 0.9.9 branch)
- fixed bug 125223 "Can't handle non-ASCII characters in URI"
- fixed bug 125080 "mailto url with raw non-ASCII subject and mail body is not working"
- fixed bug 69859 "The NS_OpenURI that returns a nsIChannel is incorrectly named"
- fixed bug 114897 "My.Yahoo - Logs you out when using Back button"
- fixed bug 127408 "cant open directory viewer to local drive"
- fixed bug 118587 "unnecessary strdup()/free() in nsJARChannel.cpp, function GetContentType()"
- fixed bug 129539 "unescaping regression from fix for bug 124042"
- fixed bug 129734 "Site name is displayed as "c" in Image Manager after just blocking site"
- triaged bugs
- Took Monday and Tuesday off.
- Landed nsGenericModule hashtable depend removal. Now this helper class can be moved into the glue library.
- Fixed a bug where async proxies could post even though they consisted of out/in-out parameters. (128560)
- Wrote a document describing what is required of XPCOM to support the Mozilla Runtime Environment.
- Triaged Mozilla 1.0 bugs.
- Working on using a flatfile to cache plugin info, instead of using the registry. bug 109739 - plugins registered in appreg each
- time they are used.
- Triaged bugs.
- worked on bug 118895 "<base> tag does not properly work when href contains a trailing '.'". marked it as works for me, as the bug is not reproducible.
- uploaded patch for bug 101731 "M098 N621 Trunk crash [@ imgContainer::FillWithColor]"
- uploaded patch for bug 107516 "Download Button at http://www.microimages.com/mix/ "
- worked on bug 121481 "forget datas". Traced the problem lies in base url not getting updated if the base href tag lies within the body tag rather the head tag.
- working on bug 97960 "First time, the Netscape feedback app started - subsequently, the entire Mozilla application closes without warning (does not crash, just terminates).". Observed that by setting mWnd to NULL (to be set as null as per mfc)in Show method of nsFilePicker, mozilla no longer crashes or terminates.
- Fixed basic authentication of inline images. Bug 62108.
- Checked in part of bug 42224 which was missed in my previous checkin.
- Fixed a problem with sending onloads when dynamically changing images. 76177.
- Have fix for topcrash bug 89595. Waiting for driver approval.
- Attended a developer conference on Thursday and Friday.
- Coordinating the preparation for the Netscape Tech. Series presentations.
- Preparing mozpoint.
- Other management activities took too much time.
- no status from gordon this week.
Plans for next week
- fix more mozilla 1.0 / nsbeta1+ bugs
- land patch to freeze nsIInputStream and nsIOutputStream
- work on patch to freeze nsIChannel
- Continue working on my nsbeta1+ bugs.
- Give nivedita more bugs to work on.
rjesup (mainly) and I (less) are working on a major improvement he
dreamed up to the way we perform reflow in response to setTimeout-driven
DHTML. Early results are _very_ promising, such as 1000%-or-more
improvement on some of our troublesome DHTML test cases. Bug 129115 has
the details, including information on pulling the branch.
This last week email@example.com have approved over 250 patches for
checkin to the 1.0 trunk. We're working hard to make approvals for good
fixes happen as quickly as possible. The more information that is
included in the approval requests the easier it is for drivers to make a
quick assessment of the request. The kind of information that should be
included in an approval request is the developer's risk analysis, the
scope of the patch, pre-checkin testing and any QA testing done. A good
request also includes a link to the bug and the names of the reviewer
and the super-reviewer. In most cases, requests that look like this:
take drivers quite a bit longer to approve than requests that look like
Please approve bug 12345. Thanks
Please approve the fix for bug 12345
<http://bugzilla.mozilla.org/show_bug.cgi?id=12345>. This fixes
a regression where the UI doesn't reflow properly if the moon is in the
wrong phase. The problem is
highly visible and the fix is medium risk. Any problems would be limited
to this one feature. I've run all of the pre-checkin tests with no
problems, I've been running with this change for several days and the QA
contact for this feature has run through all of the functional areas
that my patch might affect. The patch (attachment 1234) has r= John and