Mozilla To Implement Multiprocess Browsing

By | July 20, 2011 | 37 Comments

Mozilla To Implement Multiprocess BrowsingAccording to Christopher Blizzard, Mozilla’s Web platform director, via a blog post, Mozilla has renewed its efforts to bring multiprocess browsing to Firefox. The post stresses the key advantages that process isolation will transfer to Firefox and tackles some of the requirements for Mozilla’s implementation.

Although multiprocess browsing became a higher priority for Mozilla after the release of Firefox 4, it is still not clear when the feature might land. Current and upcoming versions of Firefox do not yet have the feature, so chances that the it might hit the public by the end of this year are slim.

Blizzard continues to mention that moving tabs into separate processes can prevent certain kinds of crashes from taking down the entire browser, enable better hardware utilization on computers with multicore processors, improve the responsiveness of the browser’s user interface, and enable superior security sandboxing.

The way in which this architectural change will impact the Firefox memory footprint is a bit less transparent. Having processes for tabs will add to the memory overhead, but it also offers some benefits. Blizzard says that using different processes will reduce memory fragmentation problems and allow the browser better predict when to release memory when tabs are closed.


About (Author Profile)


Being passionate about software, Armin joined FavBrowser.com in early 2011 and has been actively writing ever since. Having accepted the challenge, he also enjoys watching anime, indulging in good books, staying fit and healthy, and trying new things.

  • Sarjoor

    Mozilla’s separate processes project, aka Electrolysis ( https://wiki.mozilla.org/Electrolysis ) has been in the works since 2009.  I hope they finish it soon.

  • Arthur Ţiţeică

    Taking down the whole browser or just one tab. Is that really THAT important? It still means that there’s something buggy in that browser – why don’t they fix that in the first place? It’s like taking painkillers to calm the pain instead of going to the doctor to fix the root cause of the issue.

    • Sarjoor

      Yes, it is really THAT important!  Also most of the time, it is not something buggy with the browser, but with plugins or add-ons loaded by that page.  So, yes, it is really important to restrict crashes or hangs only to that certain bad tab if possible.

      That’s why IE and Chrome and the upcoming WebKit2 are so good.  Separate processes enable with increased responsiveness, parallel execution, and enhanced security.

      • Arthur Ţiţeică

        Crash with plugins it’s still a bug in browser (the way they handle plugins). The point is the bug remains but now with memory overhead. Problem solved but the issue remains if you get my drift ;)

        • catch

          Failure in software is inevitable.

          The best that can be hoped for is to minimize the failure in a way that makes it predictable to contain and recover.

          • Arthur Ţiţeică

            They keep saying that so we take it like it’s true. Adobe makes plugins focusing IE and Windows and 32 bit. Designers make sites for IE and maybe gecko. That is not failure – it’s laziness and disrespect for the end-user.

            As a concrete example: Google makes google-docs unusable in Opera. One may say that is Opera’s fault but Google isn’t the shop at the end of the street – it should/can do better. They don’t! This is not failure but corporate decision which give a crap about users.

          • Anonymous

            But it takes time to test each piece of software. The only alternative is to block or somewhat limit functionality. 

          • http://my.opera.com/rafaelluik Rafael

            Or use Web standards, that way you don’t need to actually test and retest all of the features, and fix a lot of things…

            About the multiprocess thing, I agree with Arthur.
            I prefer the browser to actually fix the bugs, and implement things like saving the text you typed on the crashed page and restore it when you get back there… For plugin bugs, make a separated process for the plugin, that’s justifiable.

          • Nobody

            this web standards is a myth and will be a myth for some years

            do a page as complex as google docs using these legendary ‘standards’.. c’mon, go show us how to do it

            a hint: even html5 is still not a finalized ‘standard’ it still does not cover everything – previous specs (html4) were of VERY low quality, did not include exception scenarios, until html5 and js and css matures THERE ARE NO STANDARDS per se. so c;mon – code according to something that does not exist!

            there are ‘industry standards’, and these are set by the biggest players: gecko and webkit – somehow they are mostly in agreement what this means. opera is always knowing better, and they suffer webpage incompatibility and thus they whine and blame it on the developers

            get to work – it seems that they’ve finaly started to fix their idiotic bugs – last desktop build contained a truckload of these (like removing INT limits on element widths/margins etc, that broke a lot of pages – opera was the only one with this issue, for years..)

          • http://my.opera.com/rafaelluik Rafael

            “do a page as complex as google docs using these legendary ‘standards’.. c’mon, go show us how to do it” In the same way they do it by using non-standards. Non-standards are a group of rules they need to follow too. The job is the same. ;)

            I’m sorry HTML5 is still being developed… The thing is, the same company behind “Google Docs” (or other “x” application) with bugs is making part of HTML5 development team, so they need to blame themselves twice…!! haha!

            I don’t think that thing of INT limits were affecting a lot of pages, it’s with absurd values anyway…

          • Arthur Ţiţeică

            Forget about google-docs for a moment. Let’s remember instant search in google and custom image for the home page. Haha!! Google – the richest company in the world can’t write another 2 lines of good javascript code or they have so much power that they do what they please?

          • Nobody

            google instant was broken because opera history.back navigation object was different than in all other browsers. google decided to drop opera support.

            masking as firefox ‘worked’ until:
            - you didnt delete a character while typing (it stopped working)
            - you didnt include non alphanumeric characters in your query (it stopped working)
            - you didnt back from the page you followed from seearch results (it took you back TWO pages in history, not one)
            etc.

            opera started to sort out their history object recently, maybe, just maybe, theyll do it good this time – given they history record, i doubt it

            and this int limit broke down google books and affected google maps, so it wasnt a ‘small deal’.

          • Arthur Ţiţeică

            :) you didn’t actually answer my question really. Did Google do what they pleased?

            You think gecko, webkit and *IE8* didn’t have bugs to be sorted out?

            You think Google would ignore 50 millions of Opera downloads (not counting Linux users) if they didn’t hope for their greater good – that is better promotion of their featureless GUI-only browser which embeds the work of others?

          • Arthur Ţiţeică

            :) you didn’t actually answer my question really. Did Google do what they pleased?

            You think gecko, webkit and *IE8* didn’t have bugs to be sorted out?

            You think Google would ignore 50 millions of Opera downloads (not counting Linux users) if they didn’t hope for their greater good – that is better promotion of their featureless GUI-only browser which embeds the work of others?

          • Nobody

            they give a crap about 1% users because it costs them more to adapt to opera then these 1% users generate revenue compared to others

            it is a business, opera is a weaker player in this game, and it is opera job to adapt and clean its own yard

            go to sitepatching blog and look for google docs related material – you’ll see that most defects they fix are OPERA OWN BUGS. get a grip

          • http://my.opera.com/rafaelluik Rafael

            LOL, actually some of them are differences because Opera implement differently, things like e.g. default value for simething is 0 or 1 that aren’t defined in the Web standards.

            Some of these “bugs” are present in WebKit and other browser too, but they put a mechanism to identify the browser and fix, while leaving Opera with no detection or “fix”.

          • Nobody

            so WHY have opera decided to use different default than all other browsers?

          • http://my.opera.com/rafaelluik Rafael

            Ok, I agree sometimes it’s a bug or not so well implemented, every browser vendor has its problems with compatibility. The fact is when Safari and Firefox behave differently the developers put a way to work with the two (by browser sniffing they take a different code for each one) and Opera is left with no code or a default for IE.

            Sometimes Opera do it for security, there’s no default value or behavior established on the standard recommendation and they choose the safer one. ( http://my.opera.com/hallvors/blog/2010/07/20/postmessage-s-targetorigin-and-security )

            Others are most browser sniffing where the website has no issue at all in Opera…
            Access and you’ll see some posts about sniffing and compatibility: http://my.opera.com/hallvors/blog/

          • Arthur Ţiţeică

            1% you say? They panic if they lose 0.01% (generally speaking – not Google). You do realize that 1% means a few hundred millions :) and you must admit that supporting IE6 and 7 usually means 10 times the work needed to support standards.

  • http://twitter.com/TiagoMorbusSa Tiago Sá

    It’s useless. I haven’t had a crash with Firefox for months and months and months. I have had crashes in the past, of course, and some of them were pretty mysterious. But I use Firefox every day for hours and a huge number of tabs, never had a single crash with Firefox 4 or Firefox 5. Chrome, on the other hand… I have. And I just use it to try it out once upon a time. IE? Crashes galore.

    I don’t know about Opera, but on my end it’s pretty stable, even if buggy in terms of rendering. It probably doesn’t have multiprocess, that’s why it doesn’t crash :)

    • Arthur Ţiţeică

      I wouldn’t call it useless it’s just the newest trend with a lot of hype around it. Browsers were almost never the problem – USERS are. Even with process separation and Linux and god knows what other security protection in place the user will still get fooled if he’s clueless about what the web means.

    • Asknobody

      Opera buggy rendering hapens because web devs don’t follow the web standards but adapt to IE and FF rendering. But now with Firebug dead, soon FF will also have “buggy” rendering :D FF don’t crach a lot but I wish it did because after some time with 50+ tabs it just kills my system and I have to kill FF manually, with crash it would be faster :D

  • http://twitter.com/bricky149 Shane Bundy

    It took Google 2 years to do the same, yet Mozilla hasn’t quite got there in the same timeframe?
    Well I guess Mozilla’s been putting effort into more important things so they haven’t wasted their time.
    There are more important things than this after all.

  • Guesty Guest

    The one thing that I think is always ignored in these situations are the variables OTHER than the browser.

    Operating system, other applications running, amount of RAM that you have installed, speed of your internet connection etc.

    I think the performance of different browsers is more noticeable if you are using them on the most clapped out, knackered old PC possible.

    Even the worst browser will perform pretty well if it’s used on a hi spec PC with a fast internet connection.

    • http://twitter.com/bricky149 Shane Bundy

      So you’re saying IE6 won’t freeze on a gaming rig?
      CHALLENGE ACCEPTED.

      • Guesty Guest

        LOL, good point. IE6 was legendary, for all the wrong reasons. I guess you can restrict my comments to the latest version of the 5 major browsers.

        I just don’t think that the companies making browsers should waste time and energy trying to get page load times to an absolute minimum, when most people on the planet can’t tell the difference between 0.6 second and 1.0 second. A second is typically the smallest unit of time that we as humans can comprehend without the aid of equipment, so as long as the pages load in a second or a second and a bit, are people really going to moan? Please note I am not advocating page load times of 15 seconds or something like that!

        The guys at Mozilla must be aware of the pros and cons of multiprocess browsing so I hope that they implement it wisely and not just to go with the flow and then have it screw up things further down the line.

        It seems like it will be 2012 before we see it. Firefox 7 onward should show improvement on memory usage but that’s doing stuff other than multiprocess.

      • Guesty Guest

        LOL, good point. IE6 was legendary, for all the wrong reasons. I guess you can restrict my comments to the latest version of the 5 major browsers.

        I just don’t think that the companies making browsers should waste time and energy trying to get page load times to an absolute minimum, when most people on the planet can’t tell the difference between 0.6 second and 1.0 second. A second is typically the smallest unit of time that we as humans can comprehend without the aid of equipment, so as long as the pages load in a second or a second and a bit, are people really going to moan? Please note I am not advocating page load times of 15 seconds or something like that!

        The guys at Mozilla must be aware of the pros and cons of multiprocess browsing so I hope that they implement it wisely and not just to go with the flow and then have it screw up things further down the line.

        It seems like it will be 2012 before we see it. Firefox 7 onward should show improvement on memory usage but that’s doing stuff other than multiprocess.

  • Senina chung

    It’s not a good idea. Multiprocess is a Double-edged sword. It will take up more Memory and CPU although it’s helpful for the freezing problem. Avant browser is a good example. Compared with the 2011 version I still like using 11.7 versions which haven’t Multiprocess.

  • Anonymous

    I will only consider Firefox when they do something like the omnibar.

    • Guesty Guest

      I am using Firefox 5.

      If you are also using Firefox 5, try this.

      Firefox Button > Options > Toolbar Layout…

      Now when the customize toolbar window pops up, you can drag the separate Google search bar into the window and then hit DONE. The search box is now gone from the UI and you will just have the address bar spanning the entire width of the page, just like in Google Chrome.

      If you type something in the address bar and hit Enter, then a Google Search will be performed just like in Google Chrome.

      For both Google Chrome and Firefox, the shortcuts for selecting the text in the address bar is Ctrl+L. That saves messing about with the mouse, and if you’re quick on the keyboard it makes perfect sense.

      So in my mind Firefox already had an omnibar. I use it all the time. It’s just that Mozilla don’t shout about it.

      Less than 10 seconds of  simple customization and you have the omnibar that you want. You can always hit Ctrl+K and be taken straight to the Google homepage anyway. If you keep the separate Google search bar visible then the shortcut to access it is Ctrl+K.

      If you want to complete the Chrome look then whilst you have the customize toolbar window open, tick the checkbox marked “Use Small Icons” and then drag the home button stop button and refresh button so that they sit in their “traditional places” between the back and forward buttons and the address bar.

      So you don’t need any add ons or to wait for Mozilla to sort this out in a future version, just 30 seconds of simple customization and Firefox basically has the same layout as Google Chrome – if that’s what you desire. If not then you can do something else with it or just leave it the way it was.

      With Chrome, you don’t get these options and the ability to play around.

      • http://www.youtube.com/user/max1cp?feature=mhum Maxim

        Yeah. With Chrome you don’t need a 300 word manual either.

        Just saying…

        • Guesty Guest

          I know I know. It’s the kind of thing I could show someone in person in less than a minute but having to explain it in writing across potential language barriers is always going to be a little more tricky.

          • Asknobody

            FF – i’ll will show how to use me, Chrome – just use me

          • http://my.opera.com/rafaelluik Rafael

            You’re being stupid, you only need to remove the search field, common!?

  • Anonymous

    Multiple memory leaks, awesome!

    • Asknobody

      FF10 – multiprocess implementation, FF11 – memory leaks get fixed, FF12 – memory leaks regarding multiprocess get fixed, FF13 – minimal requirement for FF – 1GB RAM, FF14 – all memory leaks get fixed, FF14 – FF auto-restarts after 30min of using to fight memory leaks, users rejoice, FF15 – multiprocess is off by default

    • http://my.opera.com/rafaelluik Rafael

      LOL

«
»