Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Saturday July 29 2017, @05:02AM   Printer-friendly
from the sounds-like-a-fairy-tale dept.

Submitted via IRC for Bytram

Dietrich Ayala is a developer relationist working for internet freedom at Mozilla, the non-profit makers of Firefox.

[...] I've got a Firefox profile with 1691 tabs.

[...] As you would expect, Firefox handled this profile quite poorly for a long time. I got used to multi-minute startup time, waiting 15-30 seconds for tabs from external apps to show up, and all manner of non-responsive behavio(u)r.

And then, quite recently, everything changed.

The author then describes his testing platform and admittedly simple-minded test scenario: a Macbook, time to load all 1691 tabs in HIS profile, and using Firefox versions 20, 30, 40, and 50 through 56.

The upshot? Startup time dropped from over 7 minutes to under 15 seconds. Memory usage dropped from over 2 GB to under 0.5 GB.

Source: https://metafluff.com/2017/07/21/i-am-a-tab-hoarder/


Original Submission

 
This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
  • (Score: 1, Insightful) by Anonymous Coward on Saturday July 29 2017, @06:12AM (10 children)

    by Anonymous Coward on Saturday July 29 2017, @06:12AM (#546161)

    HOWEVER, having said that, even initially loading all those tabs isn't the bad aspect to me. The worst aspects to me in order of importance:
    1. MEMORY LEAKS - Seriously, Mozilla has so many memory leaks it isn't funny. They keep blaming it on addons, but the ONLY addons I run are NoScript or uBlock/uMatrix, and even with a vanilla install you can watch memory rise just by opening a few dozen tabs, then trying to free memory with about:memory. Hint: it doesn't clean up very well, and over time it will slowly fill up more, indicating core memory handling issues in the browser.
    2. Out of memory errors/Non-memory segfaults - It segfaults! I have lost quite a bit of important state information due to this. In the former situation it should pause all tasks, inform you off the lack of memory, and request you free up more memory before continuing.
    3. Silent save errors - When out of memory, rather than throwing an error like the older netscape browsers did in the past (original and pre-FF mozilla I believe both offered out of disk space errors.) These errors can result in silently failing file downloads, silently failing complete website downloads, AND if you're really not careful silently failing bookmark errors (I think this last one was more or less fixed by the switch to sqlite, so you don't lose your whole addressbook anymore, but AFAIK it still doesn't throw an error when you attempt to make a bookmark and are out of disk space. It just pretends it saved it and if you look later it is not there, just like the 0 size filesnames you get in any of the aforementioned out of disk space error conditions.)

    All three of these complaints have been getting made to Mozilla for going on 15-18 years now. The reason they have fallen out of people's minds is because the people who cared about any of these conditions gave up on Mozilla when it didn't listen to its power users then, and those that cared moved to alternative browsers. Those that didn't, or ran into these conditions infrequently enough kept using FF, but stopped submitting bug reports they knew would never get handled.

    Mozilla was the original Lennart Poertting. Sort of like how Esound was the original Pulseaudio, even though both are conceptually NAS and probably older sound systems.

    Starting Score:    0  points
    Moderation   +1  
       Insightful=1, Total=1
    Extra 'Insightful' Modifier   0  

    Total Score:   1  
  • (Score: 2, Informative) by Anonymous Coward on Saturday July 29 2017, @06:38AM (6 children)

    by Anonymous Coward on Saturday July 29 2017, @06:38AM (#546168)

    The browser suffers from memory fragmentation above all else, that is, they allocate too many small objects and do no custom pooling (I bet some of this code goes all the way back to netscape and was never changed). The memory might be released, but due to gaps forming in the continuous memory regions that memory can't be used for newer, larger allocations once those small objects are released. They would benefit immensely from a custom memory manager designes specifically for the browser (so no lazy drop-in solutions).

    • (Score: 5, Funny) by chromas on Saturday July 29 2017, @07:19AM (3 children)

      by chromas (34) Subscriber Badge on Saturday July 29 2017, @07:19AM (#546172) Journal

      That sounds like a lot of work. They should just create a new programming language from scratch.

      • (Score: 2) by LoRdTAW on Saturday July 29 2017, @12:13PM (2 children)

        by LoRdTAW (3755) on Saturday July 29 2017, @12:13PM (#546220) Journal

        Whats a third Mozilla language going to solve?

        • (Score: 2) by chromas on Saturday July 29 2017, @07:09PM (1 child)

          by chromas (34) Subscriber Badge on Saturday July 29 2017, @07:09PM (#546377) Journal

          Well, clearly the first two didn't get the job done.

          They could pull an Opera and throw out all their effort to become yet another chrome fork.

          • (Score: 3, Funny) by LoRdTAW on Saturday July 29 2017, @09:27PM

            by LoRdTAW (3755) on Saturday July 29 2017, @09:27PM (#546428) Journal

            They could pull an Opera and throw out all their effort to become yet another chrome fork.

            Shut up and take my memory!

    • (Score: 0) by Anonymous Coward on Saturday July 29 2017, @08:26AM (1 child)

      by Anonymous Coward on Saturday July 29 2017, @08:26AM (#546188)

      It sounds like something the OS should be capable of resolving. Is there not a general purpose memory defrag that can automatically run during idle?

      • (Score: 1, Informative) by Anonymous Coward on Saturday July 29 2017, @09:30AM

        by Anonymous Coward on Saturday July 29 2017, @09:30AM (#546198)

        The OS can only do so much when someone is aggressively allocating and freeing a lot of memory. But hey, what tools the OS offers they didn't even bother to use (at least on windows --- where despite its existence they don't enable the low-fragmentation heap, never checked this in *nix but the situation is likely the same if such tools exist there).

  • (Score: 3, Interesting) by driverless on Saturday July 29 2017, @10:24AM (1 child)

    by driverless (4770) on Saturday July 29 2017, @10:24AM (#546205)

    +1. I can make Firefox die (slow down to the point where it takes 15-30s to respond to a mouse click, then eventually go into "program is not responding mode" with 100% CPU used) with five tabs or less. Just go to a site with a few animated GIFs on each page, e.g. that Pokemon-image-cartoon site my kids seem to live on (until Firefox locks up yet again). I don't care if I can load 10 million do-nothing tabs in the background, I want a browser that doesn't go catatonic with a handful of tabs loaded.

    • (Score: 0) by Anonymous Coward on Sunday July 30 2017, @03:54AM

      by Anonymous Coward on Sunday July 30 2017, @03:54AM (#546545)

      I can do that on Chrome with only one tab with youtube loaded. Chrome is such a pile of crap that I can't believe that the Fx developers want to give us a second Chrome.

  • (Score: 0) by Anonymous Coward on Saturday July 29 2017, @03:08PM

    by Anonymous Coward on Saturday July 29 2017, @03:08PM (#546274)

    1. MEMORY LEAKS - Seriously, Mozilla has so many memory leaks it isn't funny.

    Yep, but, with V56 (I've had a Nightly of V56 running for a week now) they have seriously cut down on the leaks big time. With my normal tab load (yes, too many, but not thousands) a week of usage in V53 would result in a firefox that was consuming at least 4-6G of RAM.

    With a week of usage of V56, it has gone from a start of 1.1G of RAM used to 1.6G of RAM used

    Yes, that is still 0.5G of leak, but that is much better than 5-6G of leak over the same run time. So they have made a huge improvement. Yes, they still have a long way to go, but V56 is much improved.