November 2005 - webkit-dev - lists.webkit.org

Bug life cycle
by Joost de Valk 08 Jan '06

08 Jan '06

Hi all, today, while talking on #webkit, i got two questions which both should be answered in the bug life cycle document on the WebKit page as well, in my humble opinion. To be able to do that, i'd like to get some consensus on them. The first one, brought up by Alexey Proskuryakov (ap), is this: when a bug can't be reproduced anymore but has been confirmed at one point in time, what to do with it? My answer for this would be to close it as "WORKSFORME" and not as "FIXED", since you don't know what fixed it. Ideally we would create a new resolution for this, something like "FIXED WITHOUT SOLUTION" or something like that, have you people perhaps got any thoughts on this? The second one: a bug that is fixed goes to verified, and to closed when it's in an Apple release of WebKit. However, as Mitz Pettel pointed out, we have bugs as well which are regressions from previous versions, these will, when fixed, never get into a release. I propose that we close them anyhow, since they should have a keyword which indicates that it was a regression. Everyone agree with the second? Any thoughts on the first? I'd be glad to hear from you all! Kind regards, Joost -- Joost de Valk #: AlthA on #webkit @ irc.freenode.net @: webkit(a)joostdevalk.nl W: http://www.joostdevalk.nl "The only people who find what they are looking for in life are the fault finders." - Foster's Law

4 4

WebKit Welcome Page
by Timothy Hatcher 08 Jan '06

08 Jan '06

One item we have been talking about adding to the nightly builds is a page that loads on launch. This page would thank you for testing WebKit and give you links for certain tasks to help us. This is my concept for the page. Feedback and suggestions to improve this are welcomed. http://colloquy.info/webkit-page.png — Timothy Hatcher

6 5

Safari not loading pages with domain forwarding (from GoDaddy.com)
by Hanspeter Schaub 09 Dec '05

09 Dec '05

Howdy WebKit-Developers, I've been trying to setup a small business web site which is hosted on my .Mac account. I purchased the domain http://blue-sky-prod.com from GoDaddy.com, and enabled this domain name to be forwarded to the .Mac web address. In Firefox and Explorer this works without a clitch. However, when running this in Safari (current Tiger build, or latest webkit-beta), the page comes up blank? Here is the odd thing. If I first fire up Firefox and load the page there, then the address (http://blue-sky-prod.com) loads just fine in Safari?? I tried calling GoDaddy.com support. The person there had heard of this issue a few days ago, but they didn't know what was causing this. Apparently all other browsers they try work just fine, only the Tiger+ Safari browser won't the auto-forwarded web pages. He couldn't give me any hint at how it would be for this to be resolved. He did say several times this could be an issue with Safari, in which case they couldn't do anything... Has anybody else here run into this? Is this really a Safari issue? Is this a GoDaddy.com server or domain forwarding issue? I'd appreciate any support/feedback/help on this issue. Many of our customers will be running OS X, so having Safari support is crucial. thanks, Hanspeter Schaub

5 5

New smart pointers for efficient refcounting
by Maciej Stachowiak 30 Nov '05

30 Nov '05

Hi everyone, In many places, to get memory management right, it is way more convenient and less error-prone to use a smart pointer, like SharedPtr (*). However, sometimes this leads to needless thrashing of the refcount. For example, any function that returns a newly created refcounted object should really give you a SharedPtr, to avoid leaking. But suppose that value is going to go directly into a data member that is a SharedPtr. That's going to lead to an extra ref and deref. In some spots that actually shows up on the profile. With manual refcounting you can special-case taking ownership of a newly created object, and avoid the extra ref and deref. I came up with a simple idea for how to deploy smart pointers throughout without suffering from this refcount thrash. It is inspired by the standard C++ library's auto_ptr. The basic idea is that you keep SharedPtr for most purposes; but in interfaces where you pass ownership, you use the new PassRefPtr. Passing ownership means either returning a newly allocated object (or otherwise one that the callee won't hold onto), or passing in as an argument an object the function is guaranteed to ref and hold onto. PassRefPtr works just like SharedPtr in most respects, except for assignment semantics: SharedPtr = SharedPtr // both keep a ref PassRefPtr = SharedPtr // both keep a ref SharedPtr = PassRefPtr // PassRefPtr transfers ownership and nils itself out PassRefPtr = PassRefPtr // second PassRefPtr transfers ownership and nils itself out Then you follow these rules: - local variables, global variables and data members should be a SharedPtr if they maintain a referece, or a raw pointer if they don't need to (for example a parent pointer, where the parent owns you) - function return values and parameters should be a raw pointer or a PassRefPtr (not a PassRefPtr& or const PassRefPtr&), depending on whether they transfer ownership - sometimes you might want to use a PassRefPtr for a local variable, if you plan to immediately transfer ownership Here's a test program I made that includes the PassRefPtr implementation and demonstrates use. A function getRefCounter returns a PassRefPtr which is then passed off to an Acceptor object's constructor. This runs in 2.95 seconds compared to 3.05 for this naiive SharedPtr version: So it's a measurable speedup to save those extra refs and derefs at construction time. I plan to integrate this soon. Any thoughts? Regards, Maciej * - I'd also like to rename SharedPtr to RefPtr and Shared to RefCounted, but I'm gonna stick with the SharedPtr terminology for this email.

2 2

Working HTML editor for Safari (Again)
by Fabio Invernizzi 29 Nov '05

29 Nov '05

(Hope this is not OT). My company develops a Portal application (WisePortal) and is extremely in need to support the Mac platform through Safari not only through Firefox. The main problem, and still unsolved one, is the research for a working javascript HTML editor for Safari. We've tested many of them but all seem to support only partially this browser. Even the one that states full Safari 2.0.x compatibility (Asbru Content Editor) doesn't seem to work properly (bulleted lists and numbered lists don't work). So, is there any working, but really working, Javascript HTML editor for Safari ? Thank you very much, Best regards, Fabio

1 0

How to build webkit as "fat binary"
by Dan Wood 28 Nov '05

28 Nov '05

Folks, It appears from some bugs posted on bugzilla (e.g. 5086) that WebKit is supposed to be able to build for i386 architecture, but I am not having any luck doing so. I'm trying to build WebKit as 'fat binary' -- having both i386 and Intel architectures. But I am having no luck. For instance, It's not building dftables, which is needed right off the bat: /usr/bin/gcc-4.0 -o /Volumes/BigData/build/JavaScriptCore.build/ Deployment/dftables.build/Objects-normal/i386/dftables -L/Volumes/ BigData/build/Deployment -F/Volumes/BigData/build/Deployment - filelist /Volumes/BigData/build/JavaScriptCore.build/Deployment/ dftables.build/Objects-normal/i386/dftables.LinkFileList -arch i386 - mmacosx-version-min=10.3 /usr/bin/ld: warning fat file: /usr/lib/gcc/i686-apple- darwin8/4.0.1/../../../libSystem.dylib does not contain an architecture that matches the specified -arch flag: i386 (file ignored) Am I missing something here? Is this a bug in TOT? One would think that intel compatibility would be a no-brainer at this stage in the game.... Dan Wood

2 2

Extending Safari development
by Mark Thomas 28 Nov '05

28 Nov '05

Hi, I was wondering how these guys extended Safari :- http://haoli.dnsalias.com/Saft/ With reference to adding in an extra preference tab and adding extra item to the context menu. I was investigating this and I think the path is using Text Input Managers, but couldn't see how you would go further from here or if this was the right path and then how to get the extra preference. Could you get access to the HTML source or selected src ? Thanks in advance for any pointers, Mark.

1 0

Alt tab display problem
by Mark Thomas 16 Nov '05

16 Nov '05

Hi, I thought I'd report up that if there are images are using the alt tags, and you overlay the images using the visibility to control which appears while the right image is displayed, but when show images has been switched off, the alt tag's won't appear. I've entered rdar://4330277 for this, its not hugely big for us, as we are only doing this to check that we have alt'ed the images for accessibility reasons, screen reader will read it correctly when not crashing constantly though (that's under another rdar). Thanks Mark.

3 8

A Proposal for Webkit Programmers from Karelia
by Dan Wood 16 Nov '05

16 Nov '05

Hi folks, Some of you here may have heard of Sandvox, our "iApp" website builder that is currently in Beta. We're obviously leveraging WebKit quite a bit. In fact, we're at the bleeding edge of it. And we've found some WebKit bugs that, if fixed, would really improve our app. But we are a small company, and programming the depths of WebKit is not our forté. So we came up with an idea: offer a tangible "thank you" to whomever can fix the bugs we need fixed. A win-win situation: not only will Sandvox benefit, but so will any app that uses WebKit. Full details here: <http://www.karelia.com/sandvox/a_proposal_for_webkit_programmers.html> Thanks, Dan Wood

3 2

Roll-over problem (Safari or WebKit?)
by Steven Fisher 12 Nov '05

12 Nov '05

-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 Ran into an odd problem with Safari 2.0.2 on several websites where the mouseover only works for part of some buttons. I've reported it to Apple via the reporting command, but I thought I'd share here as it may be a webkit problem. If there's any consensus on that I'd be happy to fill out a bug... Take a look at this site: http://www.melonpool.com/ The rollovers don't work very consistently on my PowerBook. When I mentioned this on the forum, someone else posted a similar experience, but only on some of their Macs. We've only got a couple data points so far, but it looks like G5s behave and G4s don't. I've captured a small (284k) movie showing the problem here: http://objectsatrest.com/my-content/melonpoolmouseover.mov - -- Steven Fisher; sdfisher(a)spamcop.net "Eagles may soar, but weasels don't get sucked into jet engines." -----BEGIN PGP SIGNATURE----- Version: GnuPG v1.4.1 (Darwin) iD8DBQFDdjg2PlY1Qa4hxhMRAssDAJ4ja6gtsZOajMRhD/XxY/qMQi+nMQCePaEB ZAa3VlH3CrmCzmBc1zcWxvI= =FN8K -----END PGP SIGNATURE-----

2 2