Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Firefox 15 plugs the add-on leaks

blog.mozilla.org

Post reply on HN
Print view View original

111–120 of 125 posts

Re: Firefox 15 plugs the add-on leaks

#111

Post by tjoff »

phaylon wrote: post #51
tjoff wrote: post #28

Earlier quoted context omitted.

If you use a lot of tabs and don't close your browser for long periods of time Chrome is vastly superior to Firefox (and all other browsers) in terms of memory use and stability. Even if you don't use any add-ons at all. I've always taken issue with Firefox, it has the broadest support and community centered around add-ons - but if you actually use any add-on the browser becomes completely useless. I've been bitten s…

I have 396 tabs in FF right now. That would be impossible to handle for Chromium on my laptop. It starts heavily sweating at 25. Does Chrom(e|ium) do lazy loading of tabs these days? That'd be the main point that makes Firefox much more usable for many tabs to me.

I have over 1000 tabs open in Chrome. I would never be able to reach 500 tabs in firefox before giving up.

Re: Firefox 15 plugs the add-on leaks

#113

Post by nnethercote »

kalid wrote: post #72
suprgeek wrote: post #18

I am very happy that the Firefox team has finally managed to lick this issue. When at every occasion I used to raise this issue, the typical Mozilla Evangelist would answer: "It is because of the add-ons". My counter to this was that the whole Raison d'être for Firefox is its add-ons community. You cannot claim that there are great benefits to this approach while simultaneously ignoring the problems caused by it. Bet…

I agree. It was extremely frustrating to see the argument "Firefox is awesome because of the add-ons. Oh, you have problems? Why are you using add-ons?". You can't have it both ways! (Also: users don't care what subsystem is causing the problem, it's your brand which suffers) That said, I'm happy for them and I'll start giving FireFox a shot again.

If someone says "don't use add-ons", they are stupid.

If someone says "you might want to be careful enabling 20 add-ons because the chances of them all being well-written and interacting nicely is low, in which case you might need to do some experimentation to see if one or two of them are causing undue problems", they are a realist.

Having said that, it is nice that we are killing off an entire class of memory leaks in one fell swoop.

Re: Firefox 15 plugs the add-on leaks

#114

Post by nnethercote »

rexreed wrote: post #25

The about:memory feature is useful to get a glimpse into what extensions and tabs are chewing up the most memory, although I haven't had much luck in FF 14 using the GC/CC to correct that. I hope that FF 15 gives the user a bit more intuitive view and control over the memory consumption on tabs and addons/extensions.

FF15 gives a lot more details about memory consumption, because "compartments" (the way the JS heap is segregated) are a lot more fine-grained. In particular, compartments created by add-ons can now be identified. But note that this only covers some add-on memory consumption; the remainder is not reported separately, and doing so is difficult.

about:memory is quite intimidating, I'm hoping to implement soon a vastly simpler version that will give a simple per-tab breakdown and be understandable by normal users.

Re: Firefox 15 plugs the add-on leaks

#115

Post by nnethercote »

antidoh wrote: post #48
jhatax wrote: post #19

Earlier quoted context omitted.

Firefox lost market share because Chrome was a leaner, faster, and better browser. I have started to hear more stories about folks switching back to Firefox because a lot of their issues with the browser have been fixed. And based on independent review sites like Tom's Hardware, there is no longer an appreciable gap in the performance of Firefox and Chrome. As regards market share, Firefox's market share would be in…

"I have started to hear more stories about folks switching back to Firefox because a lot of their issues with the browser have been fixed" There was a time, some number of releases ago (a year ago?) when FF's memory problems became too much for me, and I used Chrome(ium). I never liked it, personal preference. Finally some time within that last year FF became acceptable again, then more than acceptable, and I came ba…

Firefox 4, 5, and 6 were not good releases when it came to memory consumption. Firefox 7 had some big improvements. Firefox 8--14 all had smaller improvements. Firefox 15 has this add-on leak prevention, which is another big win. And we're still working on this :)

Re: Firefox 15 plugs the add-on leaks

#116

Post by nnethercote »

gliese1337 wrote: post #105
alainbryden wrote: post #41

> Leaky add-ons usually cause zombie compartments by holding references to DOM structures within a web page even after the page has been closed or navigated away from. This prevents Firefox from garbage-collecting the page’s compartment. If an add-on relies on persistent references to closed pages by design (such as a form data history plugin), is there a way to adapt it to this new feature to not let Firefox sever i…

Copy them. I can think of no good reason that one would ever legitimately want or need to keep references to a page that is no longer loaded. It's like returning a pointer to a local variable from a C function; you should expect it to be invalid once it's scope is gone! Copying is likely to use less total memory anyway, because it will rarely result in having to allocate new compartments for the plugins, but it will…

Each web worker gets its own compartment, and thus can be garbage collected individually.

Re: Firefox 15 plugs the add-on leaks

#117

Post by nnethercote »

mcpherrinm wrote: post #67
alttab wrote: post #64

Earlier quoted context omitted.

Care to clarify?

References are severed. There's no way to stop it. You'd have to copy out the data you want.

And in practice cutting these references hasn't caused any problems, while fixing another massive problem.

Re: Firefox 15 plugs the add-on leaks

#118

Post by nnethercote »

polyfractal wrote: post #22
twodayslate wrote: post #9

I used to be an avid Firefox user. I switched to Chrome because it was faster and had all the same extensions I used on Firefox. Why should I switch back? Chrome is still faster no?

TreeViewTab extension. Until an equivalent shows up for Chrome I will never...no, strike that, I am physically unable to switch. It pains me because I have to restart Firefox every few days or else it turns into a slow, memory-eating blackhole. But, sigh, I cannot live without that extension.

When this happens, please visit about:memory and then file a bug or email me the contents.

Re: Firefox 15 plugs the add-on leaks

#119

Post by nnethercote »

kevinSuttle wrote: post #60

I think the fault is not just the add-on devs. If anything, the add-on SDK should prevent this by default. I'm not sure when Firefox got so memory-hungry and leaky, but it's the single reason I keep jumping around to Chromium and Opera. Really, really hoping that this time is for real. It's not the first time a blog post like this has been posted by Mozilla.

Hi, I'm the lead of Mozilla's MemShrink project (https://wiki.mozilla.org/Performance/MemShrink) and the author of the linked post.

Can you tell me what you mean by prior blog posts "like this"? I remember http://blog.mozilla.com/nnethercote/2011/08/09/firefox-7-is-..., because I wrote it, but it was accurate -- FF7 did have some very large improvements.

This "Mozilla claims memory consumption improvements in every release" zombie lie is something I hear a lot, but it's just not true.

Re: Firefox 15 plugs the add-on leaks

#120

Post by nnethercote »

jfb wrote: post #68

That's a cool blog post, and kudos to the Mozilla people, but I have to wonder about the mindset of the user whose stuck with Firefox even as his setup was unusable to the point of needing to install a restart button add-on . I mean, I know from being habituated to a given routine, but that seems particularly self defeating behaviour, especially considering that decent alternatives to Firefox exist?

It's simple -- he was using NoScript and Ghostery. Those add-ons don't exist for other browsers. I can imagine that a very privacy/security-oriented user would be willing to put up with some significant inconvenience to use those add-ons.

(It's fortunate that the inconvenience has now been fixed.)

Post reply on HN