The Official Win32 20060529 [Trunk] build is out.

Peter(6) · Post by **Peter(6)** » May 28th, 2006, 5:57 am

Firefox Bugsday, every tuesday from 12-16 pdt / 19-23 utc

The Official Win32 ..... builds are out , RSS feed.
IRC channels: #firefox || #spreadfirefox || #bugs

The Official Win32 20060529 [Trunk] build is out. (Cairo build)

Previous nightly 20060528 Trunk (Cairo build)

Hourly Trunk builds:
PROMETHEUS (Lin) || ATLANTIA (Mac) || PACIFICA (Win) || Trunk Tinderbox
ARGO (Lin-Cairo) || COLUMBIA (Mac-Cairo) || GAIUS (Win-Cairo) || Cairo Trunk Tinderbox
Trunk Bonsai

Last update: 2006-05-29 -- 11:20 PDT = 18:20 UTC (all check-ins since the latest nightly included)

Resolved Fixed:

#279687[Firefox:Tabbed Browser]-Should be able to open new tab by middle-clicking "Go" button [All]
#311015[Firefox:General]-safe mode: add support for disabling all extensions [All]
#315402[Firefox:Help Documentation]-Create Accessibility section in Help Documentation [All]
#338383[Other Applications:DOM Inspector]-Implement Copy on properties in CSS Style Rules view [All]
#338443[Firefox:Tabbed Browser]-Make it easier to hit the notificationbox close button (Fitts' law) [All]
#338897[Core:Layout]-Popup window crashes browser on close via javascript [@ nsLayoutUtils::HasPseudoStyle] [Win]
#339054[Other Applications:DOM Inspector]-JavaScript object subject should be expanded when the view is loaded [All]
#339284[Core:XForms]-New permission manager broke [All]
#339310[Firefox:Disability Access]-mAccChildCount should be PRInt32 instead of PRUint32 [All]
#339311[Firefox:Disability Access]-remove unused variables in accessible source [All]
#339314[Firefox:Disability Access]-remove the unnecessary nsMaiInterface classes [Lin]
#339430[Core:Selection]-Triple click selects the wrong text (previous text) in this case [All]
#339551[Firefox:Search]-Search engine "grey text" does not match search engine and its icon [All]

Partial Landings/WIPs/Incoming:

#334018[Core:XForms]-index() with no repeat should return NaN [All]
#339488[Core:DOM]-Coverity complains that the wrong variable is null checked [Win]
#54488[Core:Widget: Mac]-[Mac] Non-draggable widgets in background windows should look disabled [Mac]
#305949[Core:XPCOM]-Stop exporting nonfrozen XPCOM functions [All]
#326466[Core:JavaScript Engine]-Implement Pythonic generators and iteration protocol support for JS1.7 [All]
#328116[Core:GFX: Thebes]-Implement Cairo-on-OSX for Thebes gfx [Mac]
#330587[Core:Widget: Cocoa]-no support for displaying sheets in Cocoa widgets [Mac]
#334650[Firefox:Disability Access]-When tabbing between links, link-selected events do not seem to be issued reliably [All]
#335540[Core:DOM]-Implement webapps session and persistent storage [All]
#336376[Core:JavaScript Engine]-Don't reserve identifiers in property identifier contexts [All]
#336378[Core:JavaScript Engine]-Support let bindings, blocks, and expressions in JS1.7 [All]
#336439[Toolkit:XUL Widgets]-Infobar not tall enough [Lin]
#336460[Firefox:Migration]-missing return warnings in profile migration [All]
#337320[Firefox:Tabbed Browser]-Session-restore service causes Ts regression [All]
#337492[Core:XPCOM]-xpcom proxies may release proxied object on random threads [All]
#337543[Core:DOM]-Use nsINode more! [All]
#337731[Core:Networking: File]-Make gnomefvs use frozen linkage [Lin]
#337774[Firefox:Safe Browsing]-improve safe browsing preferences panel [All]
#337970[Core:Layout: Misc Code]-nsHTMLDocument doesn't notify the charset resolver of the resolved character set [All]
#338263[Core:XSLT]-instance() function should return empty nodeset on invalid instances [All]
#338374[Core:GFX: Mac]-Mac doesn't support non-solid line drawing [Mac]
#338580[Core:SVG]-Improve SVG hittesting performance [Win]
#339196[Firefox:Installer]-Cleaner wizard image for NSIS installer [Win]
#339198[Core:XForms]-No Mac xforms.xpi nightly trunk build since May 4th [All]

Regressions/Annoying/Common bugs:

#270666[Firefox:Menus]-Customize toolbars causes all menus to become faded/unuseable, can't customize again until restart [Win]
#302348[Firefox:Software Update]-Update fails partial update and then hangs when update not in FTP mirrors yet [All]
#308522[Firefox:Extension/Theme Manager]-Extension check for update hang [All]
#308678[Core:JavaScript Engine]-Trunk Crash in Browser and MailNews [@ js_NewGCThing] [All]
#315699[Firefox:General]-Auto-login Cookies not always working [Win]
#323407[Firefox:History]-Back button sometimes doesn't work [All]
#324819[Core:Layout]-Fixed positioned elements now lag/flicker when scrolling (regression from bug 317375) [Win]
#324963[Core:Layout]-Menu highlight is broken/doesn't show up/not painted [Win]
#329237[Firefox:Installer]-Firefox uninstaller fails on VC8 cairo builds (R6034 C runtime library) [Win]
#330715[Core:GFX: Thebes]-Cairo incorrectly substitutes fonts and then bolds them incorrectly (incorrect fonts on Burning Edge bug lists) [Win]
#332713[Core:GFX: Win32]-CSS custom cursors crash [@ nsWindow::DataToAData][Win]
#334909[Core:DOM]-trunk shutdown topcrash [@ nsDocument::SetBaseURI] [All]
#336990[Firefox:Places]-Choosing delete on bookmarks toolbar deletes everything there indeed (incl. bookmarks toolbar folder) [Win]
#338064[Firefox:Search]-browser.formfill.enable = false, disables search suggest [All]
#339482[Firefox:General]-Window loses focus, no keyboard input accepted [Mac]
Memory leaks: list of known issues/David Baron's issues
THEBES (Cairo):list of open bugs
PLACES:list of open bugs

To copy your useragent+buildtime to clipboard, use the Nightly Tester Tools extension. (from extensionsmirror)
Get the Update Channel Selector extension to update nightly builds. (from extensionsmirror)
Get more out of the Javascript Console, see this page for Console².(Use 0.3.4.4+ or later)
If you're a daily tester then backup the valuable data in your profile!

Trunk fixes since branching for 1.5 (20050812) = ~3817

per 20060519:
Bugs ? 1.9a1 = 323
Bugs + 1.9a1 = 38
Bugs ? 1.9a2 = 66
Bugs + 1.9a2 = 1

Peter(6) · Post by **Peter(6)** » May 28th, 2006, 5:59 am

from yesterday

mossop wrote:
Peter(6) wrote:no, it's disabled because of leaks

Ah right, is there a bug # for that?

Nope, I think it's the "implement" bug that just has a comment that they disabled the feature until some leaks are fixed

Ger Teunis · Post by **Ger Teunis** » May 28th, 2006, 9:32 am

I've filed
#339516 - "Message editor on sharpdevelop forums generates 100% cpu load (hang)"
https://bugzilla.mozilla.org/show_bug.cgi?id=339516

Anyone able to reproduce/confirm?

mw22 · Post by **mw22** » May 28th, 2006, 10:24 am

Ger Teunis wrote:I've filed
#339516 - "Message editor on sharpdevelop forums generates 100% cpu load (hang)"
https://bugzilla.mozilla.org/show_bug.cgi?id=339516

Anyone able to reproduce/confirm?

I can reproduce the problem. It doesn't happen in 1.5.0.3, so this could be a regression. The bug needs a minimal testcase and a regression range.

Old Regis · Post by **Old Regis** » May 28th, 2006, 10:29 am

I made a small external tool to "VACUUM" sqlite files. (see #334010 [Firefox:Places]-Need a "Compact history database" button in options [All], VACUUM sql-statement and Storage:Performance).
The main purpose is to allow people to "clean" their sqlite files from time to time. First I tried to do that in a extension, but Firefox locks the DB and it was not possible to execute my sql request, so I made it as an external tool in C.

I provide 3 flavors: a console version, a version with MFC GUI (Windows > 2k only) and a version using a GTK GUI (Win, Linux).
I tested this on my profiles under Windows XP and Linux (Suse 10). Source, win32 binaries (statically linked except GTK GUI which is only statically linked against sqlite3 only and thus will need GTK runtime*) and various makefiles are included.
Package available here (TGZ, 852 Ko)

WARNING: this program is not intended to be run while Firefox is running. At least, it won't work at most it could corrupt your DBs !!!

* GTK runtime for windows are available as an installer here.

Uruviel · Post by **Uruviel** » May 28th, 2006, 11:09 am

Could some fellow mac person take a look at https://bugzilla.mozilla.org/show_bug.cgi?id=339482 (Window loses focus, no keyboard input accepted)?
Ow and peter do you think it's red list worthy?

Ria · Post by **Ria** » May 28th, 2006, 12:35 pm

Regis, I tried your tool (very clever that you could make it) and it cleared 0,10 MB of the bookmarks_history.sqlite file.
Probably because it was not very "dirty". I imported the bookmarks file only this morning.

supernova_00 · Post by **supernova_00** » May 28th, 2006, 12:56 pm

Regis, thanks it saved me 12%

over 300 kb

Old Mad_Overclocker · Post by **Old Mad_Overclocker** » May 28th, 2006, 1:33 pm

Nice idea regis

It saved me 1.2MB.
Can you make it clear favicons too ?

Peter(6) · Post by **Peter(6)** » May 28th, 2006, 1:37 pm

Uruviel wrote:Could some fellow mac person take a look at https://bugzilla.mozilla.org/show_bug.cgi?id=339482 (Window loses focus, no keyboard input accepted)?
Ow and peter do you think it's red list worthy?

sure Joël (add to redlist)

Old Regis · Post by **Old Regis** » May 28th, 2006, 1:44 pm

@Ria, the "dirtyness" is directly related to the amount of deletions performed on the DB. On a never cleaned bookmarks_history I had ~15%. During testing I had ~50% after deleting half of the urlclassifier.sqlite "by hand".

@supernova_00, no problem

@Mad_overclocker, I think it's possible but in another program, this one only purpose is "vacumm'ing". I will think about that ASAP (the favicon storage is really size consuming apparently).
EDIT: in the mean time feel free to use provided source. The SQL code is simple:

Code: Select all

DROP TABLE "moz_favicon";

(to be executed before a "VACUUM;" if you want to retrieve space.)
EDIT2: NB: droping the favicon table is not the clean way of doing that..

Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.9a1) Gecko/20060527 Minefield/3.0a1 ID:2006052705 [cairo]

supernova_00 · Post by **supernova_00** » May 28th, 2006, 2:07 pm

Regis, in the bug you created...pehaps a vacuu should be performed on each shutdown. for users upgrading to 2.0 this should only take a few seconds on first shutdown and each consecutive would only be a split second perhaps.

Uruviel · Post by **Uruviel** » May 28th, 2006, 2:10 pm

Regis don't you think that instead of a button to Vacuum the sqlite file a periodic clean-up would be better? This prevents UI clutter and if you do it, say, once every week on startup there wouldn't be a huge hit on performance either

Old Regis · Post by **Old Regis** » May 28th, 2006, 2:28 pm

supernova_00 wrote:Regis, in the bug you created...pehaps a vacuu should be performed on each shutdown. for users upgrading to 2.0 this should only take a few seconds on first shutdown and each consecutive would only be a split second perhaps.

I did not create any bug, #334010 was created by Brett W.

Uruviel wrote:Regis don't you think that instead of a button to Vacuum the sqlite file a periodic clean-up would be better? This prevents UI clutter and if you do it, say, once every week on startup there wouldn't be a huge hit on performance either

About your proposal regarding a vacuum on every shutdown or a vacuum at a specified interval, I disagree. If you take a look at Vacuuming_and_zero-fill or launch my program 2 times one after the other, you will see it takes a certain amount of time to do the vacuum. So IMHO, the button or menuitem or whatever permit the user to vacuum at will is better. (Note: I think there is an auto-vacuum flag in sqlite)
Anyway, feel free to comment bug 334010

supergirl260 · Post by **supergirl260** » May 28th, 2006, 2:59 pm

lol those files were getting pretty big over 10 megabyte rudueced by about 4