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

Fixes to memory footprint and garbage collection arrive in Firefox 7

extremetech.com

Post reply on HN
Print view View original

31–40 of 44 posts

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#31

Post by mbrubeck »

Silhouette wrote: post #28
rpearl wrote: post #24

Earlier quoted context omitted.

It'll ship in the nightly tomorrow... I guess Real People will be able to use it then.

And how many real people are going to switch browsers from whatever they're using now to Firefox nightlies because of some fixes to stuff that isn't broken on most other browsers anyway? If the answer is not "lots", why is this on HN today?

It'll also ship this week to around 100,000 users who are already on the Aurora channel, and in just six weeks to literally millions of users on the beta channel.

Firefox is an open source project. Part of our mission is to enable participation by the community, and that requires doing our work in the open and talking about it in public.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#32

Post by tomp »

sedev wrote: post #26

While the improvements they're promising are certainly a Good Thing, one bit jumps out at me: "introduced a significantly improved about:memory page with buttons that can manually trigger garbage collection (GC) and cycle collection (CC)... hitting these buttons repeatedly — or by hitting “Minimize memory usage”, which triggers both processes three times in a row — you can reduce Firefox 6′s memory footprint signific…

If you had read the bug report and the discussion below it, you'd know that chunks of virtual memory (VM) are not freed to the OS immediately after they are empty, but are only freed after 3 GCs. That's actually an optimization; allocating VM is expensive, and doing it just after you've freed some is plain stupid.

The problem with GCs is not that they don't actually work, it's that they are suited to a specific task. If the user is surfing the web and constantly opening/reloading pages, then GC will be ok. However, if you close most of your tabs and only leave a few open, then you might have to tell the computer that you won't be needing more memory soon and that it's welcome to free as much VM as possible. That's what those buttons are for.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#33

Post by sedev »

tomp wrote: post #32
sedev wrote: post #26

While the improvements they're promising are certainly a Good Thing, one bit jumps out at me: "introduced a significantly improved about:memory page with buttons that can manually trigger garbage collection (GC) and cycle collection (CC)... hitting these buttons repeatedly — or by hitting “Minimize memory usage”, which triggers both processes three times in a row — you can reduce Firefox 6′s memory footprint signific…

If you had read the bug report and the discussion below it, you'd know that chunks of virtual memory (VM) are not freed to the OS immediately after they are empty, but are only freed after 3 GCs. That's actually an optimization; allocating VM is expensive, and doing it just after you've freed some is plain stupid. The problem with GCs is not that they don't actually work , it's that they are suited to a specific task…

Thank you for that clarification. That has improved my understanding of the situation.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#34

Post by Silhouette »

mbrubeck wrote: post #31
Silhouette wrote: post #28

Earlier quoted context omitted.

And how many real people are going to switch browsers from whatever they're using now to Firefox nightlies because of some fixes to stuff that isn't broken on most other browsers anyway? If the answer is not "lots", why is this on HN today?

It'll also ship this week to around 100,000 users who are already on the Aurora channel, and in just six weeks to literally millions of users on the beta channel. Firefox is an open source project. Part of our mission is to enable participation by the community, and that requires doing our work in the open and talking about it in public.

> Part of our mission is to enable participation by the community, and that requires doing our work in the open and talking about it in public.

I completely respect that, and I have nothing against your philosophy whatsoever.

However, I assume that Mozilla provides those who prefer to stay up to date using the dev builds with information about significant releases via other channels. Moreover, while Firefox obviously has impressive numbers of people who do stay at the bleeding edge, is that not still only a small proportion of the total user base?

If HN winds up being that channel for every major OSS project that a few HN readers use, it will look like Slashdot a few years ago -- that is, the Slashdot that many people abandoned for a while. All I'm asking is that we leave specialised discussions about in-development features to more appropriate forums, and keep HN to mainstream announcements of wider interest (such as when an important feature actually makes it into a production build of a major product) if we're going to have feature announcements at all.

Given how often my original post has been both up- and down-voted, this seems to be a fairly polarising view, however.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#35

Post by azakai »

> Firefox 5 was all about bug stomping and the stillborn channel switcher, Firefox 6 will see the addition of lots of HTML5 and CSS3 features and more privacy controls, and Firefox 7 — at long last — will focus on memory management and performance increases.

This kind of summary is potentially misleading. It isn't like there is a preplanned 'theme' for each release, or that everybody focuses on one thing each time then switches to something else.

Mozilla constantly focuses on several things at once. So there are memory improvements in FF5 and FF6, not just FF7. It isn't as if until FF7 no one cared about memory, which the summary almost implies. (But, it's possible the memory improvements in FF7 are turning out to be bigger than previous ones.)

I realize the article was just doing a quick summary and there's nothing wrong with that. Just wanted to post this comment to avoid possible misunderstandings.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#36

Post by jes5199 »

You know, I have a lot of RAM in my laptop, but Firefox - even the latest release - still gradually slows down over the course of a single day. If I leave it running over night, my whole system becomes sluggish, every keypress takes visible time to render a character. I was pretty disappointed when I had to leave my perfectly configured Firefox behind to switch to Chrome, but the thing is just so much more stable.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#37

Post by gojomo »

Silhouette wrote: post #28
rpearl wrote: post #24

Earlier quoted context omitted.

It'll ship in the nightly tomorrow... I guess Real People will be able to use it then.

And how many real people are going to switch browsers from whatever they're using now to Firefox nightlies because of some fixes to stuff that isn't broken on most other browsers anyway? If the answer is not "lots", why is this on HN today?

I care more about 'web professionals' than 'real people' switching. This article made it likely I'll switch to the now-available early Firefox 6 versions, just to do the manual about:memory 'minimize memory usage' trick, even before Firefox 7 arrives in any form. Giant footprint and mysterious GC pauses in Firefox have been driving me to use Chrome more and more; this might reverse that trend.

Personally, this is big enough news I'm happy to see it (even in a couple different forms) on HN.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#38

Post by LukeShu »

beck5 wrote: post #9
bstar wrote: post #5

They should have been working on these issues a long time ago. I think they are far more important than many of the new features that came with v5. The only thing holding many of us back from using FF is the memory issue. Good thing the FB debugger is so awesome, because that's the only thing that keeps bringing me back at this point.

Is FB better than chrome dev tools?

IMO, yes. Maybe it's that I was on Chromium on GNU/Linux, and not Chrome proper, but I found the chrome dev tools to be buggy, and a little incomplete, like as-of-yet unpolished FB clone.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#39

Post by LukeShu »

jonknee wrote: post #15
wmeredith wrote: post #8

Earlier quoted context omitted.

I'll second that. Considering I work at an online development agency and only have 4GB of ram in my work machine, I'd say the author is a bit out of touch. This is most definitely an issue for me.

You have twice as much RAM as the people he was speaking of. I'd say 2GB is pretty low these days, glancing at Dell.com even the $499 entry level laptops are shipping with 4GB.

Wow... My laptop that I do most of my work on is packing 1GB. I guess I have had it for 4? 5? years. Of course, using things like wmii and Emacs help keep my memory usage down, but Firefox is making that harder.

Re: Fixes to memory footprint and garbage collection arrive in Firefox 7

#40

Post by bstar »

beck5 wrote: post #9
bstar wrote: post #5

They should have been working on these issues a long time ago. I think they are far more important than many of the new features that came with v5. The only thing holding many of us back from using FF is the memory issue. Good thing the FB debugger is so awesome, because that's the only thing that keeps bringing me back at this point.

Is FB better than chrome dev tools?

For me FB is better for prototypin code. You have the expanded console that allows me to write code quickly on pages that I may not have direct access to (like production pages). I feel like I'm using a text editor. Webkit debugger is much more limited in this regard.
Post reply on HN