Firefox 3.0 Preview

Brian Heater passed us a link to a PC World preview of the upcoming Firefox 3.0 release. In addition to the usual smoother UI, bug fixes, and feature updates, Firefox 3.0 will introduce several new components that should expand offline Web application functionality. The inclusion of DOM Storage, an offline execution model, and synchronization should all work together to allow for wider adoption of software like Google Apps at the end-user level. "As the breadth and depth of the competing applications expand, perhaps Microsoft's 90-percent stranglehold on the preinstalled and post-PC-purchase installation suite market will loosen, if only a bit. Then, too, if Windows Vista is any indication of what lies ahead, the company's software will continue to require ever more awe-inspiring hardware--a far cry from the light and nimble Web-based applications Mozilla engineers envision." The piece covers more than just the new functionality, of course, and should be of interest to anyone looking forward to 'Gran Paradiso.'

Comments

And it passes ACID2.

[LWATCDR]

The latest build that I got of Firefox 3 did pass ACID2. Another step forward for standards. Now if we can drag IE there.
Oh and first post.

Multithreaded UI / mthreaded Javascript please!

[thisisauniqueid]

The biggest performance hit in Firefox seems to be to do with the fact that the UI is multithreaded (as is the JS engine). Is there any chance this is going to be addressed in Firefox 3? Using a single-threaded browser in a multicore environment is painful, especially when working with many tabs at a time.

Re:Multithreaded UI / mthreaded Javascript please!

[xehonk]

Threads may not be the end all, but they might help performing several tasks at once better. I've seen firefox stop responding for a few seconds more than once because one tab was loading something. Of course my other tabs, though rendered completely, were left unusable because of the single threaded nature of firefox.

Screenshots

[homm2]

Screenshots available here.

Re:Just a Browser, Please

[brunascle]

you mean you're looking for the web, only the web?

yes, i remember the time you're talking about, the time when i considered firefox just a Galeon clone with crapping tabbing.

Re:What I hope it has

[Teckla]

1. Let me stop the damn animated gifs and flash things with the "stop" button like the old Netscape let me.

You can stop animated GIFs by pressing the Escape key. Also, if you're like me and want to stop all GIF animation entirely, hop into about:config and set "image.animation_mode" to "none".

You don't think Firefox is bloated?

[Colin Smith]

Does this mean that Firefox is getting bloaty? Not really.
    PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND
  5373 colin 15 0 246m 71m 23m S 18.9 16.3 14:08.68 firefox-bin


Seems pretty big to me. Konqueror is a fraction of that size.

Re:What I hope it has

[dvice_null]

1. Why stop them when you can totally get rid of them: Adblock plus: https://addons.mozilla.org/en-US/firefox/addon/186 5
2. http://kb.mozillazine.org/Reducing_memory_usage_-_ Firefox
3. Goto 1

Your suggestions are how ever already listed in the wish-list. The only problem is that the list contains probably a thousand feature requests, so I'm not sure when they will be implemented.
http://wiki.mozilla.org/Firefox/Feature_Brainstorm ing:User_Interface

Not PC World

[MSTCrow5429]

It's PC Magazine. Fact checking, anyone?

Re:I hope they've fixed the memory hogging.

[daeg]

You can help curb it by adjusting "browser.cache.memory.capacity" in about:config. It's in KB, so a value of 30000 means 30,000KB or roughly 29MB.

View: about:cache to see your current cache/memory status (click the links for further details).

Also note that the setting doesn't entirely stop the "runaway RAM", but it can greatly curb it. If you only view a few pages a day and use your back/forward a lot, I don't recommend changing it. However, if you, for instance, do a lot of Google searches and visit hundreds of different sites a day, dropping that setting can greatly reduce your memory usage. If you are restricted to only a few sites, your RAM shouldn't go too high anyway.

Most of them aren't leaks. Although I think there have been a few leaks regarding plugins, but I'm too lazy to go look it up.

Re:I hope they've fixed the memory hogging.

[bunratty]

And it didn't release the memory until you actually closed the program and opened it again. So you could open 12 pages, close all but 1 and it'd still be using the memory equivalent to those eleven closed pages.

Although Firefox does have memory leaks, what you're describing is far worse than any confirmed memory leak. Perhaps what you're seeing is that memory use reported by the operating system is not going down when you close tabs, but Firefox is at least releasing and reusing memory internally. If what you describe was really what most Firefox users experienced, most users would not be able to use Firefox for more than a few hours before they would have to restart it. There's no way Firefox could get the 14% usage share it has today with such a serious memory problem.

In summary, Firefox does have some memory leaks, but it doesn't leak anywhere nearly as badly as you're describing for the vast majority of users. For most users, it takes many days of use before memory leaks become readily apparent by looking at memory usage numbers alone. The real memory leaks are far more subtle than what you describe, and usually require some sort of memory leak detection tool to track down.

Re:Just a Browser, Please

[jawtheshark]

Yes, you're not alone....

I use Firefox 2.0.x at work, because it was what came out when I started working there. At home, I am faithful to the 1.5.x range. Why? Because Firfox 2.0.x is noticable slower, the interface is... let's say, not as good as the FF 1.5.x interface. Even now, when I install Firefox for someone, I'l more likely to take the 1.5.x branch than anything else.

I hope that Firefox 3 goes back to the roots...

Re:I hope they've fixed the memory hogging.

[bunratty]

You can help curb it by adjusting "browser.cache.memory.capacity" in about:config. It's in KB, so a value of 30000 means 30,000KB or roughly 29MB."

That setting for browser.cache.memory.capacity would cause Firefox 2 and Firefox 3 to consume more memory than the default setting, as long as you have less than 4 GB of RAM installed. Let's stop spreading misinformation about Firefox memory usage, please.

Re:Don't let Seamonkey die

[hritcu]

The only way you can get Seamonkey saved is by using it. Adding it to this list won't help.

Re:Just a Browser, Please

[Door in Cart]

Is there anyone other than me who wants my browser to just be a browser?

Why do I have to browse the web on something that wants to be an applications platform, an office suite, a local filesystem browser, and a dessert topping?

Agreed, but the problem dates back to the c.1995 (I think) when the HTML form-element was tacked on to the (arguably) already mature notion of a hypertext document layout rendering engine. From that moment on, the notion of what a web browser is (and what it ought to be) has become increasingly jaded. What web applications need is their own protocol, and their own scripting / markup language. But in lieu of that they continue to make their awkward home in the ill-fitting web browser -- because there's no other option. As long as web browser developers continue along this path, it not only means a bloated interface for viewing hypertext documents, but it means a grossly suboptimal environment for application development. And by all accounts there's no turning back; it's couched as innovation. Hell, you can't even stay logged into Slashdot through Lynx anymore because it relies on Javascript for some reason.

From TFA:

"Our ultimate goal is to make it so that Web applications are not discernable from any other applications running on your desktop," Mike Schroepfer, vice president of engineering for Mozilla, explained to me during a recent interview.

So by this I gather I should someday be able to run, say, Google Maps with the same speed and efficiency as, say, sed -- on my 486. Now that's ambitious.

Re:What I hope it has

[Matt Perry]

Let me stop the damn animated gifs and flash things with the "stop" button like the old Netscape let me.

You can stop animated GIFs by pressing the Escape key. Also, if you're like me and want to stop all GIF animation entirely, hop into about:config and set "image.animation_mode" to "none".

Those are good suggestions but they don't solve his problem. He wants to stop animated gifs by clicking on the stop button, just like Netscape and the old Mozilla suite used to do. His hand is already on the mouse and he doesn't want to remove it to reach for the escape key. This was a useful feature and it's a shame that it's gone.

Re:Stupid comparison after stupid comparison....

[bunratty]

the memory footprint on that thing is far, far beyond ridiculous at this point, not to mention noticibly larger than even IE7's memory requirements.

Most people find that Firefox 2 uses less memory than Internet Explorer 7:
http://scobleizer.com/2006/10/21/the-great-firefox -2-vs-ie-7-memory-test/
http://www.pcworld.com/article/id,127309-page,6-c, browsers/article.html
http://oomny.com/2006/03/24/internet-explorer-7-be ta2-and-firefox-2-alpha-memory-comparison.html
http://www.zimbra.com/blog/archives/2006/10/ie_7_a _better_b.html
http://forums.mozillazine.org/viewtopic.php?t=4685 25

Re:I hope they've fixed the memory hogging.

[Anonymous Coward]

Oh, I dunno...how about this for evidence:
http://www.google.com/search?q=firefox+memory

I clicked thru the first 10 pages or so before I got bored. That's an awful lot of hits with titles like "How to Reduce Firefox Memory Usage", and "Fixing Firefox's Memory Leak". I've tried all the about:config tweaks to try and reduce the footprint. I've spent hours digging into this crap. Firefox 2 is simply _not_ good enough in the memory arena.

Re:I hope they've fixed the memory hogging.

[SirTalon42]

Except 100 MB isn't normal for most browsers. I've had a Konqueror session open for around 17 days now (visiting lots of web pages in that time), and lets see the memory usage... 65,120 RSS! Thats Konqueror from KDE 3.5.6 on SuSE 10.1. 2 tabs are open right this second in that instance (neither very complex sites right now since I'm reading about printing to the computer's screen by using the VGA controller).

For comparison I just opened Firefox (2.0 or something, listed in about as "Gecko/20061023 SUSE/2.0-30 Firefox/2.0") and it opened to http://www.opensuse.org/ (default home page, I haven't customized or tweaked FF at all). The memory usage after letting it settle? RSS is 47,420! Lets just hope it doesn't rise too much (for comparison a newly started instance of Konqueror uses 28,888 RSS).

Now lets visit a few sites: Just /.'s front page and we're at 50,460, thats reasonable since /.'s page is much more complex than the opensuse gateway page. Open 4 articles and set it to show all comments (note, I'm not logged in so its not the javascripty version, just the pretty much static HTML) and we're at 62,516 RSS. Lets close all those new tabs and move the original to about:blank... Memory usage is now 61,852 RSS. It went down some, but didn't give back all the memory.

Now lets try the same thing with digg (without restarting Firefox): Just the main page on digg and we're to 62,296 RSS. Lets open the current top 4 articles in new tabs and see what we go to... and now we're to 69,452 RSS, lets close those tabs and move the original to about:blank again... 69,412 RSS.

Lets go back to /. and pretty much repeat the same thing and see how the memory usage goes... Just the main page and we're to 71,096 RSS. Lets open those same 4 articles and set them to -1 comments and see what we end up at... RSS is now up to 71,384, not as big of a rise as the previous time, but it did still go up (looks like maybe it replaced the old pages in cache, which would be a good thing and the slight increase could be explained by new comments).

Now lets go to about:blank then try something a bit different... RSS dropped down to 71,004 which is good. Now the different part, lets load lwn.net in the first tab, and in new tabs linux.com, sourceforge.net, planetkde.org, planet.gnome.org, and planet.mozilla.org. RSS is now to 79,432. Lets close all but the original tab and send that to about:blank. RSS is now to 77,088, it went down again which is good.

Lets try the same thing but another set of sites: original tab is amazon.com, new ones are ebay.com, bbc's site, cnn.com, google news, weather.com and wired news. The results of this one is a bit different than previous times, RSS has risen back to 77,148. Maybe we've hit a limit of how much Firefox is using? Lets close all but the original and go to about:blank again... RSS is now 75,692, dropped even more this time.

Lets go back to digg.com and see what it does... RSS is 75,962, exactly the same. Looks like its recycling some of its own memory (or loading the page entirely from cache). Now to open 4 articles. RSS has risen to 80,540. Lets close those and go to about:blank yet again. RSS is 80,308. Dropped some, but only a tiny amount. Lets go back to Amazon.com and search for 'operating systems design and implementation'. RSS is now 80,480, a slight rise. Now lets open Mr. Tanenbaum's books in a new tab (the 3 top results). RSS is now 80,552. Another tiny rise. Close those tabs and go to about:blank. RSS is now 80,488, a slight drop.

Its nice to see that in my little test the RSS didn't just skyrocket, but Firefox is still using more memory than the instance of Konqueror that has been open for 17 days (and has opened many more websites including lots of slashdot articles using the ajaxy version of the site). In case anyone is wondering: my machine has 1.25 GB of ram, and the total memory usage never passed 50% on my sy

Re:Flamebait mod was right

[Anonymous Coward]

I've never been happy with Firefox's memory footprint either.

Mozilla/5.0 (Macintosh; U; PPC Mac OS X Mach-O; en-US; rv:1.8.1.3) Gecko/20070309
Firefox/2.0.0.3

%CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
72.5 14.6 254748 114524 ?? R 1:34PM 53:30.45 /Applications/Firefox.app

1 Window, 4 tabs open:
http://maps.google.ca/
http://www.google.com/reader/view/
http://slashdot.org/
http://www.pcmag.com/

Web Developer Extension
Javascript Debugger (not open)
Netcraft Toolbar
GrApple Theme

By way of comparison, here are the same numbers for Safari (1.3.2 (v312.6)) with the same 4 tabs opened:

%CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
9.0 8.5 209612 66564 ?? S 6:43PM 0:58.24 /Applications/Safari.app/

And for Opera:
Version 9.10
Build 3588
Platform MacOS X
System 10.3.9

Same 4 tabs opened.

%CPU %MEM VSZ RSS TT STAT STARTED TIME COMMAND
3.5 11.9 258548 93216 ?? S 7:02PM 0:47.31 /Applications/Opera.app

Is that enough for you AC? Yeah, I thought so ...

(
By the time I finished testing the other browsers, ps -auxm was reporting - FF keeps my laptop toasty, burning the CPU cycles -- it's the google.

70.8 14.4 256140 113096 ?? R 1:34PM 70:11.16 /Applications/Firefox.app
)

Re:What I hope it has

[Anonymous Coward]

3. Let me stop sounds/music with the stop button.

In Vista, you can control volume separately for each application :) Not exactly what you want, but useful until the firefox dev team gets around to it. Posting anon because pointing out a benefit of Vista would get me shot.

Re:First to include the other as a plugin:

[Myen]

Eclipse won. ATF

Re:Stupid comparison after stupid comparison....

[bunratty]

one test 100 open tabs, which isn't a real-life situation, no browser should be expected to be optimized for that kind of workload.

It's not 100 tabs opened at the same time. It's 100 total tabs opened, with only 10 open at any one time. That's a completely realistic scenario, as many users open and close tabs all the time. The fact remains that most people cannot see any serious memory problem in Firefox. Some small percentage of users do, but that does not mean that most others see the same problems. In fact, the problems seem so rare that I hardly see the problem discussed on MozillaZine any more, and I haven't had a serious memory problem in Firefox demonstrated to me in over a year. The last one was a huge memory leak when viewing large image galleries, but that was fixed in an early Firefox 1.5.0.x release. Memory use is simply a non-issue for the vast majority of Firefox users in 2007.