As was anticipated, the boys and girls at Mozilla have released the final build of Firefox 3.5 today. Firefox 3.5 – originally supposed to be 3.1 – comes with many welcome improvements, chief among which is support for HTML5 audio and video tags.
The release notes have a very clear overview of the new features, so I figured I might as well just copy that to make sure I get everything:
- Available in more than 70 languages. (Get your local version!)
- Support for the HTML5
<video>and<audio>elements including native support for Ogg Theora encoded video and Vorbis encoded audio. (Try it here!)- Improved tools for controlling your private data, including a Private Browsing Mode.
- Better web application performance using the new TraceMonkey JavaScript engine.
- The ability to share your location with websites using Location Aware Browsing. (Try it here!)
- Support for native JSON, and web worker threads.
- Improvements to the Gecko layout engine, including speculative parsing for faster content rendering.
- Support for new web technologies such as: downloadable fonts, CSS media queries, new transformations and properties, JavaScript query selectors, HTML5 local storage and offline application storage,
<canvas>text, ICC profiles, and SVG transforms.
You can obviously update your installatiomn of Firefox to this latest version via the built-in updater, or your Linux distribution’s repositories will handle the matter. If you currently do not have Firefox installed, you can grab the browser from the download page for your specific platform/locale combination.
Javascript performance are not what I expect in this release.
They got a lot of time for this and it still not so good.
Have they compared it to Chrome/Safari ??
Still not so good? Try using Internet Explorer.
Competition is tough with Safari, but Firefox 3.5 just makes Internet Explorer look even more like a complete relic.
For people switching from IE, the speed increase is going to be like night and day.
Well done Mozilla on a well organised release, an excellent product (that uses less memory than any other major browser) and open-video support.
I’ll believe that when I see it. I will accept that FF has, perhaps, gotten “better”, consuming memory somewhat less rapaciously than before.
Regarding relative Javascript performance, our previous discussions have left me with the impression that you have something of an anti-agenda regarding Javascript, as a reult of screen-reader technology having lagged so.
Edited 2009-06-30 16:32 UTC
Do not talk about that which you do not understand. To find a screen reader that doesn’t support js is rare indeed, save for the pure CLI and in that case the browsers themselves don’t support JS much. Of course, whether said product makes it easy to understand what is going on as a result of js is another matter and some are better than others and they vary in how complicated they make it. The major Windows screen readers, for example, can make browsing the simplest web sites a chore whilst the Macintosh with Safari and the built-in Voiceover screen reader makes it a breeze.
But in any case, I think this is straying a bit off topic.
Thanks for the informed update on the reality of the situation. But my comment concerned my impression of how Kroc felt about the matter last time we discussed the issue.
As I posted it, I expected to hear from you. And was correct in assuming that your comment would be interesting and informative.
http://dotnetperls.com/chrome-memory
Believe it.
Bad web developers go to JavaScript first to solve their problems; JavaScript is the tool I go to once all other tools have been used properly and to their fullest. I have coded things in JavaScript years ahead of their time.
It^aEURTMs not that I hate JavaScript, I hate shoddy solutions that don^aEURTMt degrade gracefully at all.
Here’s an interesting discussion about the numbers:
http://groups.google.com/group/chromium-dev/browse_thread/thread/0f…
I would agree that a server-side approach is to be preferred. It is *usually* the better technical solution when it is feasable. What I disagree with is the idea that graceful degradation to accomodate those people who refuse to enable Javascript is somehow sacred.
If people refuse to enable javascript they should expect to miss out on stuff. If they cannot use Javascript, do to physical disability, then they should complain to their vendors of screen interpreting software.
Although, as Jacob has reported, there are, apparently, good options today.
Edited 2009-06-30 17:12 UTC
I can understand people’s desire for graceful degradation of sites, but I see Javascript as quickly becoming just as important as HTML. Web sites are no longer just about disbursing content, and often times are not just web pages, but highly interactive applications. The web isn’t just here to replace the likes of newspapers and books, but in many ways it’s here to surpass them for content distribution… not just static content, but also interacting with the user while providing information. The web used to be just about displaying content… those days are going away quickly.
As for the people that disable Javascript, they usually seem to do it because some web sites abuse it in ridiculous ways that take away from the user experience rather than add anything useful. Sometimes this happens due to developers trying to do more than the platform is ready for, but usually it’s due to very poorly written, memory leaking, CPU burning code that the authors should be ashamed of. Javascript is a good technology, but like any technology it can be abused.
“that uses less memory than any other major browser”
I’m assuming that you do not consider Opera a major browser since both Opera 9 and 10 use much less memory than FF 3.5.
that’s comparing apples with pears.
chrome uses webkit and V8 for javascript, and mozilla uses gecko.
also, optimizations are not easy to accompish.
You can say Mercedes is faster than Fiat, can’t you? Users don’t care what engine browser uses.
Well users don’t know what the browser is:
http://www.youtube.com/watch?v=o4MwTvtyrUQ
True, I ran the benchmark from here:
http://v8.googlecode.com/svn/data/benchmarks/v4/run.html
Firefox when compared to Safari is slower but with that being said, Firefox was designed to be easily portable when compared to the high optimised Javascript engine used in Safari (Squirrelfish Extreme) or Chrome (V8) which has alot of platform specific code which makes porting it to another platform (let alone architecture) not particularly easy.
With that being said, what ever optimisations that are being used one can’t avoid acknowledging that Javascript is being pushed far beyond what its original purpose was – nothing wrong with that but there are limitations as to how much performance one can squeeze out.
Is squirrelfish just safari or is it part of webkit? If so then surely it is more portable as webkit is available on more devices / platforms than firefox?
Its part of both Safari and Webkit; both V8 and squirrelfish. They aren’t offered on that many devices; maybe PowerPC, ARM and x86. There is a huge difference, however, between multiplatform (operating system) and multi-architecture. What I was referring to is architecture relating the processor specific optimisations used for Javascript. For example, if Sun were to adopt Webkit and Epiphany it would require them to add SPARC support to the Javascript engine they choose (the operating system support wouldn’t be all that difficult though).
TraceMonkey was developed in record time after both Nitro (formerly Squirrelfish/Squirrelfish Extreme) and V8 set the bar for JavaScript performance to new heights. I’m sure the Mozilla developers are content that TraceMonkey works correctly at the moment. Since Firefox 3.5 is out now I can imagine that the Mozilla developers will start to optimize TraceMonkey in ways which might cause a lot of breakage during development.
Edited 2009-06-30 16:31 UTC
For 4.0 they want to turn on TraceMonkey for the UI, which in comparison to other browsers is now slower, but this does allow for a lot of customization.
Firefox 3.5 is just fine. It quickly handles the web sites I use and it has the extensions I want.
When Google Chrome and Apple Safari get to that point, let’s be concerned about JavaScript speed, but there is much, much more to a browser than raw speed. I save more milliseconds by having extensions to help me than I could from executing benchmarks.
Do either Chrome or Safari have any adblock technologies with auto-updating blacklists? If not, I consider them to be pretty much unusable, no matter how fast their javascript is.
You might want to check the “Safari Adblock” plugin (Mac only though):
http://burgersoftware.com/en/safariadblock
Not sure if it auto-updates the blacklists but I think so.
“Pimp My Safari” has a bunch of plugins for Safari:
http://pimpmysafari.com/
Still my everyday browser is Firefox. I use Privoxy chained with Squid for added speed and security/privacy. All my web connections go through them, not just the browsers.
Edited 2009-07-01 13:08 UTC
Just switched and All-in-One Gestures doesn’t work with the new version. The Fire-Gestures doesn’t support diagonals. I’m thinking about switching back to 3.0
try Mouse Gestures Redox [1] it works great in 3.5
[1]: http://www.mousegestures.org
I’m afraid Firefox devs are making an incorrect claim on their whats new page. Safari can play <video> that is using Theora so long as the XiphQT codecs are installed, and as Safari 4 has been out for a couple of weeks + now then Firefox is not the first.
QuickTime plays the video, not Safari. Firefox is the only released browser with the video decoder built in (Chrome 3 will be the second).
Safari^aEURTMs <video> is still a plugin of sorts, and having to install XiphQT is hardly and out-of-the-box experience.
They don’t mention out of the box
besides, installing a codec is less than installing a whole browser such as firefox so it is less to do.
Even though it is using quicktime, it is still the browser supporting it + it still seems native (and seeming native is something firefox loves to try and do), for instance right click on the video you do not get a quicktime popup like you would with flash, but just the normal one. Personally I think it is better going though the native media system, why should my browser know how to decode any video format? Yes it is nicer the browser renders it rather than farming off to a plugin, but it should still use the system code, not bring in its own library. Cross platform argument? Just ship the codec for whatever framework the platform build is.
Well if you start of with the browser as default – which should be the case in the next Linux distro release, then it’s one less codec to install! Less effort for the end-user and the web developers.
Well, that’s like saying you would rather require external add-ons or codecs before your browser supports loading jpg, gif and png images. HTML5 wants video and audio to be just as natural as the current image format support – all out of the box experience and supported by all.
This also makes it much easier for web developers. They can now add more content to there websites knowing that users will be able to experience it as the web developer/designer intended.
Does it really matter whether it is Safari playing the video or Quicktime. Either way, it is transparent to the user.
In addition, which codec is more likely to be used, h.264 or Theora? I know that h.264 is patent encumbered, but when so many companies have already paid the fees and use it extensively, does it not make sense to standardize on that?
I’m not saying I don’t want Theora to win, I’m just saying that that isn’t all that likely to happen. As such, would it not make sense to standardize on something that have a much better chance of becoming a defacto standard for the web and replacing proprietary Flash or Silverlight?
I think you^aEURTMre forgetting that Firefox is the second most used browser, period. This matters to 25% of the world, and over 50% in some countries. Safari has a few meager percentage points. There will be more people using Theora by the end of this week, then people use Safari.
And there^aEURTMs bollox-all innovation going on in the Internet Explorer camp. Theora is well on it^aEURTMs way to becoming signifcantly important.
Actually, they won’t be using Theora: they will use a browser that can decode the few Theora videos around. Just as MP3 is still thriving even when far superior codecs exist (Vorbis), I don’t expect to see Theora supplant proprietary/patented codecs powered by Flash soon. Not even in 5 years.
While Firefox might matter to 25% of the world, don’t forget that other 70% taken by another browser that shall remain nameless. It might lose market share everyday, but it won’t be away soon. Likewise, most people are slow upgraders, so I don’t expect that Theora video will be vieweable (and therefore relevant) by 15% of the Internet users until a year or two.
IE users are slow upgraders. IIRC, the statistics show FF users to be much quicker upgraders. And users of truly modern browsers, based on Webkit, likely upgrade with even more alacrity.
Edited 2009-07-01 03:30 UTC
Firefox 3 overtook Firefox 2 within a week.
I don’t expect any different for Firefox 3.5 overtaking Firefox 3 … it really is such a worthwhile upgrade.
I’d like to know from where you got this information. It doesn’t surprise me that FF3 overtook FF2 so quickly in the computer-savvy community… but on a global scale?
http://www.spreadfirefox.com/en-US/worldrecord/
Everywhere except Africa, apparently.
Edited 2009-07-01 15:42 UTC
Just on this topic … Firefox 3.5 = 4 million downloads in 24 hours.
http://www.h-online.com/open/Firefox-3-5-already-downloaded-more-th…
Worldwide.
http://downloadstats.mozilla.com/
About 7.2 million to this moment as I type this.
Dailymotion is busily re-encoding videos as we speak.
http://www.reelseo.com/dailymotion-support-open-video-formats/9221/
I have heard it said that Youtube is at least looking at a similar effort.
BTW: H264 is patented, and it charges royalties, and the owners of the h264 patents have reportedly announced a plan to increase the royalties they are asking by some “massive” amount.
http://arstechnica.com/open-source/news/2009/05/google-dailymotion-…
Edited 2009-07-01 06:50 UTC
Well, that’s a single video provider among thousands. It will add Theora media on the Web, but it doesn’t make it relevant yet. Futhermore, encoding media from a degraded version (e.g. not from the source) is just a bad idea.
Now that you mentioned it, it wouldn’t surprise me if YouTube does a similar move, but only because they are owned by Google.
Nevertheless, that doesn’t change my conclusion. Until that other browser support the <video> tag and the Ogg codecs (remember they are not a part of HTML5), Theora might stay as an alternative, but definitely not as a replacement of proprietary/patented codecs.
My master thesis will deal with H.264, so I’m well aware of its status… That said, the current royalties are quite fair. Sure, they don’t fit with open/free cooperation, but it’s far from being a ripoff. That said, I didn’t heard of such plan. Link to back this? Otherwise, it sounds like FUD, like that other company with that other browser like to do.
By the way, I’m all for Theora and Vorbis… My music collection is using Vorbis and it could be interesting to work with Theora once my thesis is over. I just don’t believe the hype.
Is this authoritative enough?
http://www.mpegla.com/news/n_03-11-17_avc.html
That spells out the licensing arrangements up to the end of next year. After that, they could do anything they wanted to. Particularly, the “Internet Broadcast” bit – they charge no license fees for now, but will in the future. The concern is that they may wait until h264 video is in wide use, with lots of pre-existing content, and then start charging significant license fees for using it.
OK, in terms of licensing for actual video, this isn’t too bad at the moment, and most of the concern is speculative. But it’s still there, and isn’t a problem at all with Theora.
Of course, the per-decoder licensing arrangements do prevent either Mozilla or Opera from ever including h264 support in any product, which is why both of them are pushing Theora instead.
Edited 2009-07-01 16:13 UTC
Regardless of Mozilla’s interest … it is CLEARLY in EVRYBODY’s best interest (websites, users, encoders … just about everybody involved) to use Theora instead of h264 … the only exception would be mpegla.
Edited 2009-07-01 16:41 UTC
It doesn’t say anything about “massive” increases beyond 2010.
That’s my problem. I’m not claiming that H.264 is superior or anything like that; it’s just that speculations don’t worth much. It’s like I was speculating that Theora might infringe patents. Maybe, maybe not: until proven, it’s just speculation.
In fact, the only claim that I made is that proprietary/patented codec will stay around for a while because of that other browser.
Since Theora is patented (it’s based on VP3), I guess I can’t be wrong, no matter what happens.
Dailymotion are re-encoding to Theora from the source, not from a degraded version. Why did you imagine the latter? … you clearly know it is a bad idea so why would a large video site like Dailymotion not know that?
The royalty increases for H264 on the web will apply from 2011.
http://www.h-online.com/open/Open-source-video-codec-Ogg-Theora-hot…
From the horses own mouth:
http://www.mpegla.com/news/n_03-11-17_avc.html
“Internet broadcast (non-subscription, not title-by-title) ^aEUR“ Since this market is still developing, no royalties will be payable for internet broadcast services (non-subscription, not title-by-title) during the initial term of the license (which runs through December 31, 2010) and then shall not exceed the over-the-air free broadcast TV encoding fee during the renewal term.”
(PS: this is spin-speak for ‘from 2011 on we will charge for just streaming out video over the web, not only for the encoding function’)
Early versions of Theora encoder didn’t give competitive results … the decoder is fine though.
Thusnelda is the project to improve the Theora encoder, and that project has reached alpha release state, and it is “hot on the heels of H264”. Currently the difference (for the same bitrate and compression factor) is just 2db.
http://web.mit.edu/xiphmont/Public/theora/demo7.html
This is not hype. Its fact. The first two pictures of Theora 1.0 vs Thusnelda frames say it all. Furthermore, optimisation has only just begun. There is room for much improvement still, so no doubt the Theora encoder will overtake H264 quality and compression sometime soon.
Edited 2009-07-01 16:36 UTC
So they kept the source of over 300K videos for which they don’t own the copyrights? Well, if that is the case, then I stand corrected. It’s just a little bit odd to me.
That is your interpretation. Maybe they will, maybe not. Right now, it’s too early to call “massive” increases with the available information.
Well, that is a no-brainer. The main challenge in video compression is encoding…
That page isn’t hype; the way you promote it is. Two frames and statistics for a specific clip (Akiyo) using an alpha version? Definitely not enough to conclude that this new version of Theora is getting better. This is not an attack on the developers: I’m sure they are fully aware that tweaking may lead to regressions.
The H.264 standard covers the decoder and the bitstream. It does not cover the encoder. Hence, most H.264 implementations got lots of room for improvements, just like Theora. Don’t expect them to stay idle, especially when they’ve got industry-wide support!
Sorry, but you are living in cloud cuckoo land if you believe that.
After 2010 there will be a fee for anyone to stream h264 video over the internet. A “pay-per-play” fee, applied as if it were a television broadcast. The fee per transmission itself will be tiny … but remember that a television station broadcasts a given video stream once (or at most a few times) to a large audience, whereas a website sends and resends the stream many thousands of times to individual recipients.
This will mount up to become a prohibitive cost for video websites.
This is the reason why sites like Dailymotion, Wikimedia and a few others are right now moving their encoding to Theora, and sites like YouTube are considering doing exactly the same.
http://www.nabble.com/Google-Chrome-to-support-Ogg-Theora-video-nat…
There is a defensive patent on the original VP3 codec upon which Theora is based. Theora itself now holds a valid, irrevocable license to it, and Theora is onwards-licensed to anyone via a BSD-style license.
http://www.theora.org/faq/#24
There have been no challenges to Theora. I realise that this doesn’t “prove” a negative, but right now there is at least one thing that is definitely proven … h264 definitely IS patent encumbered, and the owners of the patent definitely ARE going to charge for its use.
Edited 2009-07-01 23:40 UTC
Well, I’d rather live there than in paranoia land.
Your conclusion is right if the premise is. Did the MPEGLA (representing the licensors) clearly stated their intentions on enforcing those fees? It would make sense, but unless they stated their position, we are in speculation land.
I was already aware of the position of Theora. For some reason, you always seem to assume that the OP is a complete ignorant and/or an idiot. At least that’s my impression by reading your comments over the months. That is irritating.
See above for your second part (that is a red herring, btw).
Since we are going in circles, I won’t argue further. My point is that proprietary/patented codecs won’t go away soon. Even with Google around.
They should be given every encouragement to do so.
Another way of looking at it is that Firefox is used more than any single version of IE.
Yet another way of looking at it is that without doubt in a matter of a week or so there will be more Firefox 3.5 users than IE8 users, and IE8 has been out for many weeks now.
Yet another way of looking at it is that Firefox is supported on more platforms (and hence devices) than any browser (with the possible exception of Opera).
No, because who is going to pay the licensing fees for all the other platforms (excluding Apple and Windows)? What about OS’s like Linux, *BSD, ReactOS, BEOS, OS/2, ….. Theora is free to all, so everybody can enjoy the internet as it was intended by the web designers. Theora improves with encoding quality at every release, and for web quality (which is much lower than when you rip your favorite DVD collection) that is not a issue anyway. Also Theora has better streaming support as far as I know.
Page load times and interaction with complex web pages feels tangibly faster than 3.0. Interaction speed anecdotally seems comparable to Chrome.
Glad they are making improvements, but 3.5 still just looks sort of the same. It’s hard for me to get excited about a video tag. If anything, it seems like this release was too little, too late. Chrome and Safari get 99% use by me at this point on Windows and Leopard respectively.
I’m curious if this is just going to be one of the downsides of Firefox – being “completely” open source may mean it will lag behind other available browsers in terms of features.
What are you babbling on about, both Chrome and Safaris engines are ^aEUR~open source^aEURTM (Webkit), the backend is where the features come from.
I disagree.
At this point, the rendering engines are all pretty good and steadily updated, so the real features come from UI (Firefox is kind of boring), usability (Firefox is again kind of boring), and security (pretty much neck and neck, although Chrome is going for the win with the sandbox mode). Plugins may let Firefox catch up a bit, but I don’t know… I feel like any time I use Firefox, I am going back in time.
I don’t know what to make of the acid3 test. Who is better? Mozilla or webkit? Does Mozilla ignore acid3 and just try to be as standards compliant as possible? Does webkit specifically target acid3?
Not at all, Mozilla has been working on Acid3 as well as other things. Reaching 100% is important, but its not the be-all-and-end-all of web standards. The test was created around existing buggy support and missing support between the browsers.
As of today, 3.5 moves Firefox from 72% to 93%, which I think is very acceptable.
I can’t agree more! 93% is excellent considering all the crappy designed web pages out there. The web designers need to start catching up before the 93% will even become an issue – at which point Firefox will be at 100% already.
In terms of acceptability, Firefox’s 93% acid3 score stomps all over IE8’s 20% score, thats for sure.
Yes, the WebKit developers put in quite a lot of effort a while back on achieving that 100% on Acid3 – it was a bit of a race between them and Opera, I recall.
Mozilla never got into that race because at the time, they were in code freeze for FF3.0 – afterwards, they don’t seem to have worried about it much, and just picked up points here and there from work on standards conformance, rather than explicitly targeting Acid3. So here they are at 93%, which if not yet perfect, is certainly good enough for most purposes.
Allot of people are annoyed buy the inclusion of location aware browsing in Firefox 3.5. In many cases it does not jive with their privacy concerns.
See this bug:
https://bugzilla.mozilla.org/show_bug.cgi?id=501546
Such a big release, but so silent. I don’t see any marketing and other stuff.
http://www.spreadfirefox.com/
http://www.mozilla.com/en-US/firefox/features/
http://blog.mozilla.com/blog/2009/06/01/show-us-your-speed/
Edited 2009-07-01 07:02 UTC
Mozilla are refreshing folks! I don’t really get the people bitter about the release, it’s a present for you after all. A strong brand, transparent development and an excellent browser (plus it most likely runs on the system of your choosing). It was fun to watch the number of blockers going down, marketing team getting ready, QA team ripping the builds apart, etc and poof, the release is here. It is a living and very well organized phenomenon, appealing for geeks and non-geeks as well.
Mozilla is also very solid in graphic design, their webpages, logos etc. are lightyears stronger than that of the competition, and this is one crucial element of success.
3.5 is perceivably smoother at some places. Complex pages are faster as others noted. The bookmarks manager still has that sluggishness even when dealing with a small number of items. Some other performance bugs are dealt with. There are more privacy-related settings in the UI, hidden by default. Wording and other small things make this release clearer for the newbie. The last tab behaviour is alien. Tab tearing is very welcome.
(My gf prefers Opera because it doesn’t reload the previous page when hitting Back. This would be nice to have.)
I hope major webpages are going to pick up Theora and friends soon, I really can’t stand Flash boxes, especially on Linux.
Congratulations and a cake to the Mozilla team, looking forward to Namoroka!
Firefox behaves the same way, I think. A proper web site will provide the validity period of a response via the Expires HTTP header. Firefox won’t reload the page upon pressing Back unless it’s expired. Firefox 2 (iirc) introduced the “Back-Forward Cache”, which speeds this process up significantly.
It really does make you wonder why anyone would complain, doesn’t it?
So, what is it REALLY like, as seen by technical reviewers who don’t have a barrow to push?
http://arstechnica.com/open-source/reviews/2009/06/hands-on-firefox…
Surely, in anybody’s book, 5,000 enhancements can’t be wrong?
Surely it is a good thing to have technology that can deliver rich web content to users on ANY platform?
Firefox 3.5 is a good (nay, great) thing for (almost) everybody.
Edited 2009-07-01 10:08 UTC
I was not very hopeful that this release of FF would be much better than the last one, despite the comical version number bump.
I typically run with 150+ tabs open, and when I say “typically” I mean that I never go below this number and atypically go as high as 500 or so. This means that I see, feel and am bitten by memory issues sooner, more and more often than most FF users. If you thought FF was slow for *you*, try running FF 24×7 with 200 tabs open.
Crash recovery (session saver and descendants) has been critical to my FF experience because it meant tracking this huge state across the frequent crashes got easier. With the introduction of FF 3.0.x I found that I went from crashing often to crashing infrequently, which was lovely. But that didn’t solve the number one problem.
The biggest issues I’ve seen are memory usage and CPU usage. FF3.5 was supposed to be better with JS efficiency, so I was hoping for a modest reduction in CPU load (in other words, perhaps a constant 20% CPU usage by FF instead of a constant 30% usage). What I found instead was far, far better: Memory usage has gotten better and stayed that way.
FF3.0 was nice because it no longer leaked like a sieve, but it still ate gobs of my RAM. When I open FF and my tabs reload I tend to go up to about 800M of RAM used, then that goes up as the session stays open. FF3 has not had the familiar eat-everything-and-die we remember from older versions, but it would still eat sometimes as much as 1.5G of RAM.
I opened 3.5 I watched my resource usage and… it ate 550M of RAM, then stopped. Right now, a good while later, it’s using 645M. This is unheard of, for me, crazy-low. What’s more, the average CPU load hasn’t gotten better by a third, it’s gotten better by a half or more. I’m seeing no more than 15% load from FF3.5 when idling.