Submitted by Kimmo Kinnunen wrote yesterday on the GTK+-WebCore developer mailing list that he has imported the Safari 2.0 WebCore branch into GTK+-WebCore. “This means that from the webcore/javascriptcore part, the code is mostly
the same as in current Safari. So if there are any crashes, they’re not
from webcore/javascriptcore part of the codebase with very high
probability, rather my code.”
So now KHTML has travelled from the KDE libraries to Apple to become a part of Webcore and is now being ported to Gnome. I think that a software library in common between the two main Linux desktops and OSX is pretty significant. Much as people bellyache about fragmentation, seems like there’s convergence going on where it matters.
As Dasher42 posted, this is quite significant from a purely software development perspective, but as a web developer it’s also pretty interesting: one of the better light-weight free (beer/speech) rendering engines out there (‘better’ in that it balances standards-compliance and being lightweight very well) is getting a broader appeal.
this really is fantastic, gnome deserves a better html renderer. and don’t say we have gecko- that is so complex and difficult to develop with.
I’m so impressed with kde and gnome they are working together a more. maybe KDE should take the plunge and implement gconf or elektra. Gnome and KDE will benefit greatly when some GPL’d java comes along.
Gcko is great and Gecko > KHTML. Gecko is more than a lame Redering Engine such as WebCore. Gecko is very easy to develop… just learn it.
Im so impressed with kde and gnome they are working together a more.
Somehow it’s fitting that the first KDE tech I can remember Gnome adopting came through Apple, Gnome’s inspiration for simple and easy. Now I wonder if pressure will increase for KDE to drop KHTML and use Webcore. It’s totally shallow, but I’d prefer KDE to use a KDE lib instead of an Apple one even if the ancestry is the same
Seriously though, the cross DE stuff is nice. DBUS, Webcore, etc. Yay!
maybe KDE should take the plunge and implement gconf
I’m not so sold on this one. Maybe you can convince me? People are scared of a “registry” sometimes, though obviously gconf isn’t a big binary blob, but rather XML. I’m still not convinced that XML is the greatest way to store settings (as opposed to a more easily human readable flatfile). Instant Apply is nice, but one doesn’t change settings all that often. Leaving aside my preference for flatfiles, what’s the benefit of two DEs with mostly unrelated programs using the same config backend? Enlighten!
And yeah, GPLd Java will be nice. I’ve messed about a bit with Trolltech’s Qt bindings. It was a bit annoying relearning stuff from Richard Dale’s earlier bindings for KDE3, but I managed to get an app ported over.
“I’m not so sold on this one. Maybe you can convince me? People are scared of a “registry” sometimes, though obviously gconf isn’t a big binary blob, but rather XML. I’m still not convinced that XML is the greatest way to store settings (as opposed to a more easily human readable flatfile). Instant Apply is nice, but one doesn’t change settings all that often. Leaving aside my preference for flatfiles, what’s the benefit of two DEs with mostly unrelated programs using the same config backend? Enlighten!”
I agree with you, XML is not always the best way to store configuration. Also, I certainly am not a huge fan of gconf. Although I’ve never really come up with a better idea for an alternative myself.
Nonetheless, I think the answer to your question about unrelated programs is that several linux users, myself included, tend to mix and match applications accross DE’s. Personally, while I do try and maintain consistancy of my applications, it’s more important to me to use the best o0f breed software. So while I run Gnome I use K3B because, in my opinion, it’s leaps and bounds above Gnome-Baker.
In terms of configuration, different programss may suit different configuration formats better. But, sometimes it’s better to think of the greater good of the whole, ie the complete ecosystem of Open Source software. It’s better if all things may be configured in the same way. This will allow DE agnostic configuration schemes that can be managed centrally. Unfortunately this may hurt more for some applications than others but it is for the greater good of the entire DE and even will promote cross DE pollination that will help to improve everything. Whether a gconf style XML format is the best option is another matter. But I think at some point the community will decide to suck it up and adopt a single configuration standard for all software.
It may be a pipe ream but it’s my pipe dream
G
Funny how some ideas never die out…
GConf ist not limited to XMl. It’s not a “registry”, it doesn’t even remotely compare, except for the hierarchical namespace.
GConf is only an API. You can store all GConf-values in any kind of storage, be it a libsql-database or even a LDAP server. You can even use a flatfile-backend. All you have to do is write it.
Edited 2007-04-08 15:23
good, safari is a fast, and very strickt browser!
The less rendering engines (less browsers doesn’t matter) I have to debug for, the better. Now if only Opera would adopt KHTML…
Why should they do that? Opera already has a better renderer than KTHML/WebCore/Safari.
It would make no sense for them to adopt an inferior one.
It would if they want their market share to go above 1%. Web developers have to deal all the quirks of Trident, Gecko and possibly KHTML/WebKit if they have access to it; debugging for Presto (Opera) on top of that to (with yet another JavaScript engine as well) is not worth the effort for minimal gains in users. Not even Prototype support Opera officially, they have too much hacking on their hands as it is.
That may be true, but I’ve been using Opera for a while, and I don’t think they’re too worried about desktop market share, to be honest.
They’ll take it, but from what I’ve seen they’re more focused on phones and PDAs, which is where they actually make money. The desktop browser is still great and it gets a lot of their attention, but they’re just not too concerned about taking market share from the other desktop browsers
Besides that, there’s not too much hacking needed to get pages working in Opera. If a page works in Firefox, and passes validation, it usually works in Opera with minimal hacking.
But that’s just my experience, and I’m a big Opera fan, so I’m a little biased.
Edited 2007-04-07 13:28
If you avoid the newest fancy stuff or avoid IE-specific code Opera, Firefox, Konqueror, osb-browser (Flower) and most other browsers will render pages quite fine.
XHTML 1.0 + CSS 2 usually just works. Well, don’t use PNG’s if you want it to look good with Inter Explorer 6
Opera’s flaw no.1 is that it feels very foreign on all platforms. It is poorly integrated in Windows and poorly integrated in Linux. And it doesn’t have all the nice extensions Firefox has.
However:
– it’s starts faster than Firefox (but still too slow)
– has fewer problems with frames and flash (Firefox has this annoying bug when clicking on a link all or most frames on the page is opened in individual windows – and closing one of them crashes Firefox – it also happens with Epiphany, Galeon and Kazekahaze when compiled against Firefox) – however Opera still crashes on occasion when visiting sites with excessive use of Flash (solution: annihilate the webmaster!)
– uses slightly less memory (but still way too high memory consumption)
– Opera doesn’t become unresponsive to same degree as Firefox when loading heavy sites (like ekstrabladet.dk or something ugly bloat like that – the proper solution would be to annihilate the webmaster).
Opera can be used but is not much better than Firefox, and the lack of extensions makes it irrelevant.
In order for me to consider another browser it has to support HTML, XHTML, CSS1+2+3, Java, JavaScript, and Flash – and have a low memory consumption and start fast.
If osb-browser had support for Java and Flash, Firefox would be unmerged instantly.
I think that foreign feeling depends on the default setup of the platform and the browser. For me defaults are foreign 🙂 Since I’m using KDE with Reinhardtstyle i just installed the “Simple” theme for Opera, activated use of native colorschemes there and that was it. The only thing which nags me is that it automatically opens the download tab when i’m downloading something.
Solution is known for years. It is in Opera’s preferences. But Preferences in Opera was from start designed by some geek, starting from version 1.0 – strange things located into strange places.
Shortly – look somewhere in “History” (not sure if they didn’t renamed/replaced it again), find Cache section and switch off “Automatic” for RAM cache size.
Set it to amount you desire.
Mem-hungriness stops with that.
I don’t think the lack of extensions is an issue, considering that Opera has just about everything a decked-out copy of Firefox has built into it. Basically, it suffers from most of the same flaws as Firefox, though not in such an exaggerated fashion.
LOL
Opera is almost as bad as Firefox and Gecko. Opera is slow, bloated, unstable and extremely memoryhungry. I tried to replace Firefox with Opera, but the slowness, the lack of stability, the memory consumption and lack of – for me – useful extensions meant I left it behind.
It has a reasonable renderer but it is no match for WebCore/KHTML.
I’m looking forward to Epiphany on GTK+-WebCore.
What exactly do you see as “extremely memoryhungry” in Opera!? And what is slow? For me it is the other way around. Firefox is slower and uses more memory, even without useful extensions installed. Ok, not that much, only 10 MB. Subjective speed seems to me at least double in Opera in startup and rendering of most sites compared against a bare Firefox. Stability? No real concern. It doesn’t crash that often. Maybe this is a question of general System setup?
Edited 2007-04-07 12:48
Opera is slow in regard to loading the browser itself. It is clearly faster than Firefox. Again Firefox is almost as slow as OpenOffice (e.g. 2.2 seconds vs. 3.0 seconds for OpenOffice 2.1.0).
osb-browser loads in 0.3 seconds
(but does not have support for Java and Flash).
Opera uses slightly less memory than Firefox but it still uses 3-4 times more memory than it should for the tasks it is performing. Firefox is worse yes. But Firefox has a lot of good extensions which Opera doesn’t. I am willing to sacrifice those extensions if I got a fast and light browser instead. Opera is not such a beast. It’s a dinosaur like Firefox. Slightly smaller, slightly faster, but still a dinosaur.
Opera generally doesn’t crash – except for Flash-ridden sites. Firefox is even worse then (a bug that has been around since 1.0.x – and is present even in Firefox 2.0.0.3 (showed up after upgrading to Flash-9)).
It is most likely a problem with Flash rather than the browsers, but a buggy plug-in shouldn’t crash the browser.
I suppose you can prove all of this too with hard facts? Sorry, but Opera is anything has a reputation for being fast, not bloated, and NOT memory hungry. Whereas FireFox has a reputation for being slow, bloated, and memory hungry. A *provable* one as well. Didn’t see you see Thom’s post a few months ago about constant problems with FireFox and how he started using Opera instead?
Depends on how you define memoryhungry and so on.
Compared with my criteria Opera is big and bloated and slow. But faster than Firefox. I’ve written that several times.
Opera is bad, Firefox is worse.
What is “provable” numbers? Does output from exmap qualify as such? And output from time (used to measure loadtimes)?
I appreciate your wish for documentation, but I need to know what you consider documentation.
Do you have any particular wishes for websites to visit and number of tabs open in the browsers?
No, because exmap is Linux specific, so I can’t comment on that.
However, output from Solaris prstat gives me:
20971 username 131M 62M sleep 59 0 0:00:06 2.5% firefox-bin/10
8048 swalker 84M 73M sleep 49 0 0:00:30 0.7% opera/1
When I have the following URLs open:
my gmail inbox
http://www.osnews.com/reply.php?news_id=17646&comment_id=228480
http://www.google.com/
http://www.mozilla.com/en-US/firefox/2.0.0.3/releasenotes/#contribu…
This is all immediately after starting Opera9/FireFox 2.0.0.3 respectively.
Notice the size of FireFox is about 50 mb bigger…
I don’t care about default settings being bada, or page caching mechanisms, or anything like that. The point is it uses more, and FireFox often bloats to an unreasonable size.
See Adam’s posts about this too:
http://www.firsttube.com/read/Youre-Killing-Me-Firefox
http://www.firsttube.com/read/A-Final-Word-on-the-Firefox-Fiasco
Considering your claims started with none, and Opera has a *reputation* of being the opposite of what you said…
No, because exmap is Linux specific, so I can’t comment on that.
However, output from Solaris prstat gives me:
You can’t comment on output from exmap because it is Linux specific, ehh? But I can comment on output from prstat despite it is Solaris specific, double ehh?
Okay, I’ll be nice
– can we agree not to hang to much on the numbers but instead look on the tendencies? (We’ll have to do it that way, since we cannot compare the numbers directly – you know, apples and oranges :p )
The tendency from the output of prstat fits the tendency from the output of exmap, AFAICT. I have posted the output in a second reply to you.
The output from prstat proves my point: Opera is bad and bloated and Firefox is terribly bad and bloated. You of course interprete the results as Opera being lightweight and Firefox being heavy. I interprete the results as Opera being heavy and Firefox really heavy.
A more correct interpretion would probably me that Opera is medium weight and Firefox heavy weight. But that still makes Opera bloated IMHO. Less bloated but still bloated. And I haven’t claimed anything else.
Considering your claims started with none, and Opera has a *reputation* of being the opposite of what you said…
That’s a weird statement. It almost looks like an attempt to insult me, but then – English isn’t my native language, so I’ll give you the benefit of doubt
Opera’s “reputation” is irrelevant. Some claim Opera is light weight. Your numbers prove otherwise. A reputation can be fitting or not so fitting. It is not so fitting for Opera
Reputation doesn’t matter – facts do.
No, it doesn’t prove your point. Especially since we have never agreed upon a definition of “bad and bloated.” Since you have never defined what bloated is in exact terms or how Opera is bloated, I cannot accept your claims.
I’ll give an example you can appreciate. Many users complained when software started supporting multiple languages because it meant switching to using libraries and encoding that use more memory.
Many of the users were angry because they perceived it as “bloat” since the only reason the program was using more memory was to be able to support languages they didn’t use.
Yet, since English is not your “first language,” you know that support for multiple languages is not bloat and is very important.
In the same way, Opera and FireFox undeniably provide functionality that WebCore does not, however, that comes at the cost of more memory.
Maybe the problem is the terminology. Perhaps a better thing to say instead of “Opera is bloated” would be to say “I think Opera uses too much memory.” At that point, you would have to do a feature-by-feature comparison of Opera to WebCore to determine if you could fairly compare Opera and WebCore’s memory usage.
Obviously, the more functionality a piece of software provides, the more memory it will use. The two are not always directly linked, but in general, I think it’s a safe thing to say.
Yes, because prstat gives numbers that are comparable to Linux’s top. exmap is a Linux kernel module so leaves me with no reasonable point of comparison without understanding exactly what it does.
DISCLAIMER:
Look at the tendencies – and not at the exact number. It cannot be directly compared with output from prstat on Solaris. It would be a comparison between apples and oranges and very non-scientific. The output can only be used to establish one or more tendencies.
Memory Consumption – measured with exmap:
Websites visited:
http://www.hattrick.org/ I don’t use gmail
http://osnews.com/permalink.php?news_id=17646&comment_id=228480
http://www.google.com/
http://www.mozilla.com/en-US/firefox/2.0.0.3/releasenotes/#contribu…
Firefox 2.0.0.3
Effective Resident (K): 44313,86 Effective mapped is identical
Writable (K): 26512,00
Virtual Memory (K): 197456,00
Sole Mapped (K): 42540,00
Mapped (K): 52780,00 Resident is identical
Opera 9.10
Effective Resident (K): 29966,74 Effective mapped is identical
Writable (K): 11484,00
Virtual Memory (K): 125776,00
Sole Mapped (K): 28932,00
Mapped (K): 34016,00 Resident is identical
Kazehakase 0.4.5/GTK+-WebCore SVN
Effective Resident (K): 18006,32 Effective mapped is identical
Writable (K): 10828,00
Virtual Memory (K): 104096,00
Sole Mapped (K): 16368,00
Mapped (K): 25892,00 Resident is identical
— and just for the fun of it —
X.org
Effective Resident (K): 23924,65 Effective mapped is identical
Writable (K): 27532,00
Virtual Memory (K): 45588,00
Sole Mapped (K): 21196,00
Mapped (K): 40476,00 Resident is identical
DISCLAIMER:
Look at the tendencies – and not at the exact number. It cannot be directly compared with output from prstat on Solaris. It would be a comparison between apples and oranges and very non-scientific. The output can only be used to establish one or more tendencies.
However, your basic comparisons don’t take other things into account.
For example, WebCore does not provide all of the compatibility and functionality that FireFox and Opera both provide. For example, Opera has a built-in Mail client. Khazekhase does not. Is it fair to compare an application that is a Mail client and web-browser in memory usage to one that is just a web browser? Isn’t that like comparing the memory usage of Photoshop to Paint Shop Pro, when Paint Shop Prod doesn’t share the same set of functionality?
Do you have Java/Flash/other plugins loaded? That would change the effective memory usage for FireFox and Opera.
Finally, of course Kazehakase is going to be smaller — when it doesn’t even begin to provide half the functionality that FireFox and Opera provide.
For example, it doesn’t appear to support plugins, which means no Java, Flash, or other support.
I also fail to see how your numbers justify your claims that Opera’s memory usage is high or somehow wrong. If anything, your numbers seem to prove that FireFox is a memory hog, as I claimed, and that Opera uses a reasonable amount of memory for the task at hand.
A fairer comparison would be the Gecko engine alone embedded somewhere, the Opera engine alone embedded somehow, and then WebCore.
Knowing that Opera and FireFox are full applications while this webcore based one is not, tilts the numbers unfairly to its advantage.
Adam’s posts are not so important. Using Task Manager as a “proof” discredits the analysis, though I agree with him.
And he compares apples with oranges in regard to memory usage on Windows and Mac OS X, respectively. But the tendencies are still valid IMHO.
My experience is that all Gecko-based browsers have memory issues – especially Mozilla Suite/Seamonkey and Firefox (XUL).
I don’t care about default settings being bada, or page caching mechanisms, or anything like that. The point is it uses more, and FireFox often bloats to an unreasonable size.
I agree about issues with default settings and caching mechanisms. And I agree that Firefox often bloats to an unreasonable size. I will go as far as stating Firefox always does that. It’s just a matter of letting it run. OTOH the size of Firefox is unreasonable even on a fresh start.
The same goes for Opera but less so.
Either you are a moron, or completely misinformed, Opera is no match for khtml, simply no match..
Step 1:
Open a terminal and run svn co https://gtk-webcore.svn.sourceforge.net/svnroot/gtk-webcore gtk-webcore
Step 2:
Go to gtk-webcore/trunk/JavaScriptCore and run autogen.sh && configure && make && make install
Step 3:
Go to gtk-webcore/trunk/NRCore and run autogen.sh && configure && make && make install
Step 4:
Go to gtk-webcore/trunk/NRCit and run autogen.sh && configure && make && make install
Step 5:
gtk-webcore/trunk/osb-browser and run autogen.sh && configure && make && make install
Step 6:
Type osb-browser to run browser.
Note: It may or may not compile. (Works for me though.)
EDIT: Replaced : with and run.
Edited 2007-04-07 01:21
Now this is what I call progress.
(Loads pages twice as fast as Epiphany, using a third the memory and taking less than half as long to start up. Renders things beautifully. There are definitely bugs, but it still makes anything Gecko look like a ponderous dinosaur.)
KHTML gone right around in a big circle. Its a great engine and I’m glad to hear the gnome folks will be able to benefit from it, though in a somewhat different and modified form of webcore.
This is pretty good news, GTK+-WebCore is actually really portable.
This isn’t Gnome only, This allows one to make a browser that isn’t reliant on KDE libs or w/e.
The codebase is actually quite lighter then Gecko based browsers.
All ya need is glib, GDK, Pango, Gtk+, libglade-2, libcurl..
🙂
Takes then entire WebKit and makes the ObjC calls reference equivalent Frameworks for GNUstep and releases a Browser for GNUstep.
i hope that gecko is phased out of ephiphany- webcore is much better,
The problem is not ephiphany. Whatever people says, gecko is a great rendering engine, and in firefox 3 it’s only becoming more powerful and more efficient. Firefox 3 eats 10 MB less than Firefox 2 when opening a google docs document, for example, and javascript is way faster (and it’ll get even faster when they start using the JIT-oriented JS machine donated by Adobe). It doesn’t matters how bad is gecko, if they continue improving it at this rate,
The problem is not gecko. The problem is that crap called “gtkhtml”. GTK has his *own* implementation of a rendering engine, and that’s what webkit aims to replace. gtkhtml is not gecko, it’s not webcore, it doesn’t support JS or CSS AFAIK. It’s what evolution, gaim, sylpheed, etc; uses. It’s so crappy that ephiphany has to use gecko, because gtkhtml isn’t useful for rendering anything beyond simple html tags.
By using webkit, gnome will be able to use a REAL browser engine in all its apps. That’s what matters.
Edited 2007-04-07 14:15
You are right.
Epiphany is not the problem. It is just a browser shell.
Firefox 3 is not available yet, and as such is irrelevant.
GtkHTML isn’t the problem for Epiphany. As you mention Epiphany doesn’t use GtkHTML but Gecko. And that’s the problem with Epiphany. Gecko. Gecko.
GtkHTML is not crap. You just don’t know what GtkHTML is meant for.
As a browser engine GtkHTML fails miserably, but that’s because GtkHTML is not a browser engine. It is a simple HTML rendering/editing library designed for applications that require very simple and lightweight HTML functionality. It is not meant for an ultimate HTML browser.
Use the right tool for the right job. GtkHTML for simple HYML-documents, and WebCore/Gecko/KHTML for complex websites.
WebKit doesn’t aim to replace GtkHTML. WebKit (GTK+-WebCore) aims to replace Gecko.
Gecko (even in Firefox 3) is big, bloated, slow and an embarrassment for FLOSS.
that why kde is better
kde produce better software than gnome…
k3b
konqueror
koffice
amarok
how long to gnome community to take other kde project?
I’d love this to replace Firefox, get rid of Firefox; they had their chance, their source is still of a bloated (30+mb download), and the speed is atrocious; and worse still, it is completely out of place in the GNOME desktop.
Someone at Firefox needs to realise this; stop trying to made a damn operating system, and just write a damn browser that does just that, allows me to browse, without the need for crap like XUL, which no bugger is using, either now or anytime in the near future.
Someone at Firefox needs to realise this; stop trying to made a damn operating system, and just write a damn browser that does just that, allows me to browse, without the need for crap like XUL, which no bugger is using, either now or anytime in the near future.
You know, many of the future web standards that are being written rigth now focus into making possible a sort of web-OS. Firefox is already implementing many of them. BTW, Firefox 3 rendering is blazing fast
Yup, 1/3 of the memory usage on my machine stems from Firefox. Firefox is the browser equivalent of Vista.
Yup, 1/3 of the memory usage on my machine stems from Firefox. Firefox is the browser equivalent of Vista.
I’ve got 2 gigs of Ram and Firefox is:
PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
15 0 241m 41m 19m S 0.0 2.1 0:01.58 firefox-bin
For me, Opera for Linux used to be fast up to version 9 (betas for 9.0 were fast, too). I’ve been happily using it for years. But then suddenly, 9.0 was a disaster and nothing has improved since then. I’ve tried it on several different distros with all major versions and some unofficial beta releases, including 9.20, but on all of them, Opera is simply the slowest piece of software I’ve ever seen. Not with one or two tabs, but more than 10-15 tabs equals superslow everything in Opera (the whole program is unbelievably slow). Something got terribly broken in 9.0. But then, I don’t think switching to a different rendering engine in Opera would help.
Hm. Can’t see this slowing down of Opera after v9.x. Using Opera 9.10 build 521 linked dynamically against qt and lesstiff on Archlinux under KDE.
For me it is Opera->Firefox->IEs4Linux from http://www.tatanka.com.br as last resort for sites which work with neither of the former two. Konqueror i’m using not that much. Why? I would like to. But it just doesn’t work good enough too often. Depends on what sort of sites one needs to visit i guess. Many hardwaremanufacturers sites don’t even work under Firefox. That’s why i have this IEs4Linux as fallback.
Am I the only one who hates Safari? It crashes on me several times a week, sometimes so hard it completely wipes its own settings. Yeah, Firefox isn’t as fast, but at least its stable.
You may be interested in trying Webkit, which is Safari with a later version of the HTML engine. There are a lot of fixes and particularly also speedups. It’s not always as stable, as it’s bleeding edge, but there are new builds every day.
Webkit is available here:
http://webkit.org/
Links please?
or are people simply speculating that next Epiphany will be based on this etc?
Kazehakase-0.4.5 can be built against the newest GTK+-WebCore.
Besides that there is a reference browser in the GTK+-WebCore project – it is called osb-browser.
Just download the sources (or make a checkout from SVN) and compile it. You’ll need the sources anyway if you want Kazehakase to build against GTK+-WebCore
There has been some experimental builds with Epiphany using GTK+-WebCore.
Kazehakase-0.4.5: http://kazehakase.sourceforge.jp/
GTK+-WebCore: http://gtk-webcore.sourceforge.net/
For SVN checkout look at my earlier post.
Thanks. I just compiled it and took some screenshots of the rendering for everyone to see. =)
Though it crashed a few times on me I guess it’s still experimental stage but not too bad. =)
Even when launching for the first time this launched in less than 0.5 of a second. I have compiled it with only webcore engine support though.
http://members.lycos.co.uk/siimo2005/gtk-webcore/apple.png
http://members.lycos.co.uk/siimo2005/gtk-webcore/microsoft.png
http://members.lycos.co.uk/siimo2005/gtk-webcore/mozilla.png
http://members.lycos.co.uk/siimo2005/gtk-webcore/osnews.png
http://members.lycos.co.uk/siimo2005/gtk-webcore/useragent.png
Looks really good. And yes. GTK+-WebCore is a snappy loading browser engine. The reference browser starts in 0.3 seconds, so I’d expect other browser front ends to start rather snappy as well. Compare it to 2.2 seconds for Firefox
You just made me compile kazehakase with webcore support
yeah the osb browser is pretty snappy. I was actually surprised how many features are actually in the browser considering its a reference browser.
Ooohh.. this is nice.
Kazehakase using GTK+-WebCore starts 4 times faster than Kazehakase using Gecko, 5 times faster than Opera, Galeon and Epiphany – aaaaaaaaand… more than 7! times faster than Firefox 2.0.0.3
– and almost as fast as osb-browser.
Now, that’s a browser
It still crashes here and there, but the SVN version of GTK+-WebCore is clearly better than the old WebCore version, though it still claims to be WebKit 146.1 (it is 418.x if I’m right – that’s at least the WebKit version for Safari 2).
Now we need KWIQ fixed, so layout is better (line spacing, widget placement and stuff like that), and support for Java and Flash
(and of course stabiling the beast).
I’m all for ditching Gecko in Gnome
Just for what it’s worth, does your CPU go bonkers if you drag a terminal over the Kazehakase window? I’d kind of like to know if that’s a Webcore issue or a bug in osb-browser.
Naah.. nothing special here. It jumps a bit when moving the terminal window, but no more than what is normal for applications running on X.
However, dragging a window over the Firefox window.. that makes my cpu go somewhat bonkers.
Of course – if I drag the terminal window around fast the cpu will go nuts – but that is true for all window combinations.
I hope they subsequently move to work this into evolution (the mail/calendar/etc client)
The GTK-HTML renderer in that sucks royally. (not that I really want a fully fledged renderer, but I kind need it to test html emails for sites I develop for clients.
Having basic working CSS support and whatnot would be really nice.
I’d like to see a dozen viable Web Browsers. For developing Client-side solutions, as long as these browsers meet the standards than they can choose whatever HTML/JS combo they want.
If they pass the Acid Tests and more then my code will work.
I follow your directions posted above for compiling gtk+-webcore. Everything compiled and ran fine. I tested out osb-browser with a couple of sites (slashdot, osnews, planet.gnome) and the rendering was quite good-albeit there were some minor css issues. It is blazlingly fast and extremely tiny.
I then followed up on your post where you mentioned kazehakase. I downloaded the source to the newest version. I was worried about having to tell it where to find the gtk+webcore libraries-but no problems encountered. ./autogen –prefix=/usr, ./configure –prefix=/usr, make, make install…
And wow. All I can say is wow. Kazehakase is extremely fast-startup in less than 1 second(first time ever started). Rendering is equally quick(pauses until most of code for page is already downloaded then swoosh-drawing the page in one go-again quite fast). checked out the sites listed above and minor css rendering issues gone(no idea why). Kazehakase appears to share code form the old Galeon browser. Although the whole is pre-alpha it works good- the gtk+-webkit port is new and the gui seems to have been around a while-although it is not very pretty it is very functional. Thanks I now have a new alternative browser
Just figured out that kazehakase is using gecko-strangely enough when i watched it compile it mentioned both gecko and gtk+webcore….now i will look if it is possible to get it to build only using webcore….
you have to edit the file ~/.kazehakase/kzrc
In that file add layout_engine=gtk+-webcore to the section [Global]
thanks dylansmrjones
I am now submitting this post via kazehakase using the gtk+-webcore renderer. I also tried to disable building gecko-but to know avail. Ok now i see the same slight rendering bugs I saw in osb-browser(lol makes sense). Mostly just placement issues with css. And rendering is not quite as speedy as firefox-but still amazing given how young this is.
Do you know anything about future plans of kazehakase ? Obviously this is a first-go at incorporating gk+-webcore for rendering for this browser-you can feel the disjunct between the UI and the rendering subsystem(ie. you can’t change font size in the UI using gtk+webcore,you can use simple zoom, but when you go under preferences and manually choose the font size this is ignored for gtk+-webcore, whereas it works fine with gecko.
Am I correct in thinking this code reminds me of Galeon? Unfortuantely the web page of the author has very little information…..
It would indeed be cool if someone were to take this code and fully integrate it….A Galeon-like browser using gtk+-webcore could vie with epiphany-and definitely should be used in things like yelp and other document-viewing tools(if small enough could be used to render webpages for nautilus(preview, thumbnail etc.))
… torrent link?
*LOL*
It is not big enough to warrant for a torrent