“A lot of people complain about the Firefox ‘memory leak(s)’. All versions of Firefox no doubt leak memory – it is a common problem with software this complicated. We look to fix the issues where we can. David Baron and others have done a huge amount of excellent work in this area. What I think many people are talking about however with Firefox 1.5 is not really a memory leak at all. It is in fact a feature.”
It’s interesting how Opera/Safari can have a fast tabbed interface too and not have this “feature^H^H^HMemory issue”.
At least you can go to about:config and turn off this “feature”.
Opera does have an in-memory cache, and had it for a while. That’s why it renders the page so fast after going ‘back’. Then the feature gets added to Firefox and “oh noes, memory leakz0rs!!oneone.”
Noticed the back feature is not that fast anymore.
Definitely not as fast as OPERA.
Seems like it’s been slowed down since it’s last couple of versions.
Last couple of versions? It was added in Firefox 1.5. Nothing changed about it in Firefox 1.5.0.1. What other versions are you talking about?
Perhaps this is because Opera/Safari doesn’t “cut corners” – this is the expression Ben Goodger used when siding with Apple in the old “safari development not open enough” debate. In his opinion, instead of focusing on code perfection, khtml developers should cut some corners to make khtml render more pages. I guess we see now the effects of this kind of development philosophy (isn’t this why we usually bash Microsoft btw?). http://apple.slashdot.org/article.pl?sid=05/05/13/1227247
I have switched away from ff a year ago – both memory use and CPU reqs are ridiculous on any platform, compared to Opera (which I use on WinXP) and Konqueror, which renders most pages just fine (which I use on PC-BSD).
If you take a look at the comments below his blog – guess which question is avoided consistently… how does Opera provide the same feature without any “memory leaks.” (Or I can ask the same about Konqi: how does it manage to render pages so fast with only 5Mb of cache using less cpu cycles and less memory).
Edited 2006-02-16 13:08
Because of this “feature” I’m sending this post from Konqueror… Firefox is slow, firefox is bloated – firefox users have been saying that for a long time and nothing changed. If it isn’t a memory leak it is simply a bad design (which is even worse).
Firefox has a preference browser.sessionhistory.max_total_viewers which by default is set to -1. …
You can set it to 0 to disable the feature, but your page load performance will suffer.
Possibly the most negative reasonable assertion is that the default is daft.
Well, I used to really love Firefox, but I find myself using Opera or Konqueror more often now. I started using Firefox when version 0.7 came out, and though it would crash often it was more fun surfing the web than with 1.5, though to be fair 1.5.0.1’s memory handling has improved.
What really keeps me from switching completely to another browser are a bunch of extensions that I grew accustomed to and that make browsing more fun. But then, it’s pretty sad when third-party additions are all that make an application valuable.
What really bothers me is that those memory issues have been known for a long, long time and the developers chose to ignore them. Just browse the bug reports and see how many reports have been closed because problems were not acknowledged and downright denied.
Anyway, the last two memory leak fixes have slightly improved my browsing experience and since they made better memory handling a priority now I am looking forward to the upcoming releases. I’d really love to love Firefox again.
But then, it’s pretty sad when third-party additions are all that make an application valuable.
Actually, I thought that the whole point of Firefox was basically to strip out most of the features and add them back as extensions so that you could customize it as much as you wanted, and most importantly leave out the stuff you didn’t need.
The modular approach is quite laudable indeed, but the core application still should meet some demands that I actually deem to be self-evident, such as: being lightweight, being stable, using a computer’s resources in a sensible manner and generally not acting like it’s the only application running.
No, what I meant with the statement you quoted was that I think an application should be valuable without needing to rely on third-party additions. The extensions aren’t everything. A browser should still run fast, render pages correctly, support standards, be secure and stable. Firefox does some of these things well and fails at others, which unfortunately has become quite frustrating for thousands of users. And if it weren’t for those many useful extensions, many would have switched to another browser a long time ago.
Yes, that’s correct. But a good basic browsing experience is the basic feature firefox is supposed to ship with.
Instant back/forward is a part of that.
Actually, I thought that the whole point of Firefox was basically to strip out most of the features and add them back as extensions so that you could customize it as much as you wanted, and most importantly leave out the stuff you didn’t need.
And what exactly were the features that were stripped out and put into extensions?
Most of the extensions I’ve seen were created by third-parties who don’t even have an adequate understanding of the Firefox architecture and therefore create poor performing extensions.
And yet they find their way into official mozilla site. After insatlling OpenPGP I couldn’t uninstall without experiencing most unusual and unacceptable bahaviour (tab with message info would get whole screen high). Try this: I have 512MB RAM + 2GB swap partition – it was devoured (whole of it) by firefox & mplayer-plugin (which by the way works flowlessly on Opera) in some 30min. – result ? System (masterpiece Ubuntu Breeze 2.6.12-10-k7) rendered unusable (even getting to command line took some 15 min. – after 30 min. I gave up and restarted the machne).
If I had to pick another browser, it would be Mozilla itself, not this clumsy, unresponsive slouch firefox.
Finally I must say I just don’t get the extensions idea at all. One has to find them all after firefox upgrade. Opera, which is browser, mail client, IRC chat client, RSS news reader and now even has bittorrent client…and widgets, with so _many_ features and customizalbility takes _only_ 4MB in installation binary. How is it, that firefox – which has only the browser funcionality has to take 8.1MB in installation binary (bare applicatoin without extensions) is beyond my comprehension.
Now you can mod me down. I know, I know – provocation – that’s what it is…
“But then, it’s pretty sad when third-party additions are all that make an application valuable. ”
Let me ask you a question : What would you add to a browser to make it an excellent piece of software, of course being able to render pages is already a given….
I know ppl have touched on it already and it’s more just a side note. A browser is a boring application to begin with. The extensions that are most useful to people are those that allow your interaction with the browsing environment to be as painless as possible. It difficult to include a bunch of inbuilt features that are useful to everyone. The one size fits all doesnt work here and here’s why.
People have very personal and specific information requirements while they are browsing. I am sure if you compared the extensions that everyone using FF on OSNEWS you will find little in common.
Even ad blocking, pop up extensions aren’t extensions that everyone uses.
The blog clearly states that what is commonly perceived as a memory leak is actually a cache feature. The title here, however, seems to suggest that the developers have used the “it’s not a bug, it’s a feature” excuse.
The quotation marks around the word ‘Feature’ make it even more provocative (and wrong). May I ask why this was done? It would have been more correct to quote ‘memory leak’ instead, since that is how it’s written in the blog.
He is saying that what is commonly perceived as a memory leak is in fact a feature. He LITERALLY uses the word ‘feature’.
“What I think many people are talking about however with Firefox 1.5 is not really a memory leak at all. It is in fact a feature.”
HENCE it got quotation marks.
The title still suggests that the memory leak is a ‘feature’, while the blog entry speaks of a feature that is incorrectly perceived as a memory leak. You could have quoted an entire sentence instead of just one word.
“The blog clearly states that what is commonly perceived as a memory leak is actually a cache feature.”
Cache is a set area, not an expanding one.
The memory use keeps on going up hence the implementation of the feature is buggy. It causes a memory leak and the memory never seems to be returned, and forever expands.
There is no way you can cut it, saying that the behaviour is by design. It’s a poorly implemented feature.
The memory use keeps on going up hence the implementation of the feature is buggy.
No, there are a set number of pages, so the only way this could cause memory use to go up is if the new pages are all taking up more memory than the ones they are replacing.
If the memory keeps going up then it is a memory leak and has nothing to do with this feature.
Bad design doesn’t imply memory leak.
While I won’t argue that Firefox takes up a good bit of memory, I really don’t get how people can call it slow. Firefox, IE, Mozilla, Safari, Camino, Epiphany, Konqueror, aside from the user interface and subtle differences in page rendering, they all feel about the same to me. In fact I just ran a few tests right now just to make sure I’m not crazy.
Start time to a blank page:
Firefox 1.5.0.1: 3 seconds
Camino 1.0: 2 seconds
Safari 1.3.2: 1 seconds
Opera 8.51: 3 seconds
IE 5.2: crashed
Mozilla 1.7.12: 10 seconds
Konqueror 3.1.4: 2 seconds
Time to load osnews.com:
Firefox: 3 seconds
Camino: 2 seconds
Safari: 5 seconds
Opera: 2 seconds
IE: N/A
Mozilla 2 seconds
Konqueror: 10 seconds
About the only things that stand out here are Mozilla’s load time and Konqueror’s page load time. Everything else happens so fast that my times probably aren’t even accurate. But that just proves my point. Everything else runs equally fast.
Now granted I ran this all on a Mac so things may be slighty different elsewhere, but then again I’m no stranger to Linux, Solaris, Windows, etc, and I’ve never noticed things being much different on any of those systems.
That’s just because you have a lot of RAM. I have on my laptop 128 Mo, what should I do, buy a new computer only to browse websites ?
Tristan Nitot explained all this stuff before on his blog (not a bug, a feature). The only thing I don’t get is : why then Opera uses much less memory ?
I have 512 megs, which I certainly wouldn’t call excessive these days. But sure, my results might be different with only 128 megs of ram.
I have tested Firefox 1.04 on the old Laptop Winbook XL running a Pentium Pro 166Mhzwith only 64 MB. Initial loading take 10 seconds but accessing webpages took roughly 3-5 seconds.
Granted it might run slow with more than 10 open tabs but my father only use it to check e-mails so two tabs is enough.
128MB of RAM is a pathetic amount to have in any machine this day in age…
How did you test page load time? I think, that it has to be tested on local server, because you can’t blame browser for net slowdowns.
Browser load test should be made at least 10 times. How did you test it?
But see, that’s the thing. Nothing’s really slow. And sure I could get all scientific about this and average out 100 start times and whatnot, but I really don’t feel up to it. Feel free though.
What you need to also consider is other things besides for page load time:
– Switching between tabs. In Opera, this is instantaneous. Firefox there is a delay
– Typing a url in the address field. There is a slight lag on this in Firefox.
I’m sure there are a few other subtle things. But people notice these, and it gives off a huge perception of being slow. And well, those things ARE slow compared to some competitors. You may think it’s stupid and minor, but if it’s the kind of thing you do a lot and it lags, some people can’t stand that (like me).
Yeah well, I don’t really see it. Switching between tabs is pretty much instantaneous here on my end, no matter what the browser is. Maybe my system is just too speedy and has too much ram to make for a good test platform. I dunno.
My system is 2.2Ghz with 1gb of memory, so it’s not at all slow. However, I do notice a very slight delay in switching tabs. It’s probably half a second or under, but I notice these things, even if most people don’t.
And for the type lagging in the address bar.. that’s much more noticable.
These things shouldn’t bug me, nor do they affect anything, but they do bug me, and mostly because the same product from a different vendor doesn’t do it. There’s something in the back of my mind that asks “Why can’t they make these simple things faster? If they can’t do the simple things, why should I trust the browser to be faster than the other browsers?”
It’s kind of illogical, but eh.
What you need to also consider is other things besides for page load time:
– Switching between tabs. In Opera, this is instantaneous. Firefox there is a delay
– Typing a url in the address field. There is a slight lag on this in Firefox.
Sorry, I just don’t see that on my system. Firefox is just as fast as Opera. It would be interesting to see a study that checks how fast/slow various browsers are at various tasks for a variety of systems. Determine what makes the most difference in performance – CPU, RAM, harddrive, type of net connection, etc..
There is stuff out there, but it always gets accused of bias.
What about opening a new tab? That has always been very noticably slower for me and everyone I know. It’s still relatively fast, but it’s definitely not instant like in Opera. IE7 is even worse.
New windows are definitely slow on Firefox. I’d say it takes a solid half to full second on my Athlon 1800.
Which really isn’t that bad when you stop to think about it. It’s not great, but it’s tolerable.
But it makes the product feel less polished.
Ahem, seconds try (Firefox crashed on the first one, heh).
I really think the time it takes to open a new firefox window is awful. I know the actual time to layout the new window for the WM and get the gtk widgets put in isn’t nearly that long!
It’s standable, barely. I imagine it’s a lot nicer on newer machines.
Let’s just say, I don’t open new windows very often on my 700 Celeron laptop!
These things are slow, but Opera does not properly queue them. That is, click several links in a row in FF to open new tabs. Itakes a few seconds, but all are opened. In Opera, one or two will be, then it gets busy, and I get the scroll cursor.
I think FF needs to work on its speed and RAM use, but it has many parts of the UI right that others do not, making for a more enjoyable experience.
P.S. I even managed to talk about Opera and FF and not invoke Fitts. I’m proud of myself ;-P.
I’ve never had a problem with that. Sounds like a bug maybe?
I’m using 9.0tp2, so who knows.
what version are you using? Its never happened to me, I’m even trying it right now, no problem.
8.51 for Win32. It makes it difficult to use for forums or dense news sites. If you wait until the page is actually loading, it’s fine to click another link. I typically go to 20+ at a time, then skim over them and close tabs.
In FF, it takes forever to load them, but they all load. In Opera, I get a few tabs in, it stays busy, and that’s it until they are all loading.
I imagine ad servers don’t help the situation.
It’s irrelevant, because the issue here is memory leak. So the real problem arises when you open tons of tabs and then even after closing most of them your pc is still crawling, then you have to exit FF, or even kill it.
Perhaps you failed to notice the people complaining about Firefox performance at the beginning of this thread?
My experience tell me that seamonkey are faster than firefox, with the same engine.I think the diference are the pressure to make new versions in firefox (like some linux distros).
Edited 2006-02-15 16:58
Thom doesn’t bother to mention what the feature is,
and chops the quote to deliberately make Firefox look bad. It’s a feature…and let’s not bother to say what the feature. The feature is an in-memory cache so that ‘back’ happens more quickly. The point is is that it’s not a “memory leak” at all, and he’s calling page caching a feature, not a “memory leak” a feature. I’d be surprised if you even knew what a memory leak is, Thom.
A reasonable title would be “Firefox 1.5 page caching uses more memory”, not some sensationalist BS such as Firefox Memory leak a ‘feature’.
No he doesn’t. It was quoted on slashdot in the same idiotic way. He tries to make it surprising.
The actuall article title isn’t any better.
Sensationalism is valid for article titles. It’s silly, but it’s pretty traditional journalism. People who only read headlines deserve to have a completely wrong view of the world.
And I don’t read /., nor care about it much. Just because /. does it doesn’t mean it’s “valid”, what ever you mean by that. Care to explain why sensationalism in article titles valid? Because people who only read titles are stupid? Good strawman, but try again.
And no, the actual article title says “About the Firefox “memory leak”, while that’s not the best title, it certainly doesn’t go calling a memory leak a feature.
It’s entirely incorrect in the same way OSNew’s title is… Article titles are like this on every newspaper in the country, and so I don’t know why you expect more from a meta news site.
No, it’s not entirely incorrect. Taken literally, his title is true. His blog post is about the so-called Firefox “memory leak”.
But Thom’s title, “Firefox Memory Leak a ‘Feature'” is totally incorrect because the blog post says that it’s a feature, NOT a memory leak, not that it is a memory leak AND it’s a feature.
And so what if everyone does it? Does that automatically make it a good idea to do it?
Give ’em a friggin’ break and quit nitpicking.
The fact that everyone does it may or may not make it ok; but it means that you should take it into context when interpretting it: Sensational titles are expected. They’re trying to get you excited, in some way, about the boredom that is ordinary life.
Truly sensationalist journalism is when the article following continues the bad pattern of just trying to get you excited and not informing you of true, accurate, unabated facts.
If Thom’s title is interpretted with the given article as context: It says “Firefox Memory Leak a Feature.” All he has to do to make that non sensationalist in any way is quote “Memory Leak.”
Would you feel better if he quoted that? Or should he just put “Firefox Rocks, This Feature Concerning Cached Pages Is Not a Memory Leak: You See, Memory Leaks Occur When…” Or maybe he should go with “An Article Concerning A Supposed Memory Leak In Firefox.”
See, the problem I think you’re having here is that you’re assuming titles are factual. They’re not. They’re catchy short things to stick in people’s minds later that will help them remember the rest of the topic. They’re eye catching. They’re supposed to bring the reader in. They need to be short, of course.
They’re an abstraction. A creative, catchy abstraction. And guess what, they’re a _very_ lossy abstraction.
I’m going to give Thom the benefit of the doubt and say that he didn’t realize how bad that title was because English isn’t his first language… (It isn’t, is it?) The alternative is that he’s just trying to drive up traffic to get more ad revenue.
And this is somehow different than the same force that drives newspaper sales?
Sensationalism is a journalist’s best friend. I think you’d be hard pressed to find a news venue that didn’t prey on the credulous and gullibility of their audience.
I’m going to give Thom the benefit of the doubt and say that he didn’t realize how bad that title was because English isn’t his first language…
In this case, is it better to let a native speaking English editor to review the title before submitting it?
In this case, is it better to let a native speaking English editor to review the title before submitting it?
Or people could just stop with the pointless Thom-bashing. That got old real quick when it was still Eugenia-bashing and has become even more annoying now.
Every damn article the same nonsense over again, just quit it already. Why don’t people just mail the guy if they think they’ve got a real grievance ?
Edited 2006-02-15 18:38
Umm, yeah, email him and have him ignore it. That’s a really good idea (TM).
Actually, the title is misleading. It should be titled:
Ben Goodger: Firefox ‘Memory Leak’ a Feature to match the actual quote from Ben. Ben Goodger: Firefox Memory Leak a ‘Feature’ is completely different. You can call “sensasionalism” but misquoting someone hinder the reporter.
There goes another example of the Open Sauce fanatics. I laughed at you.
Please refrain from posting on the web. Think of the children.
So this feature sucks >32MB *per page*… what the heck are they doing, storing a screenshot of the page? (the term ‘rendered document’ seems to suggest so, but IMO a browser should really be able to render a document quickly without displaying just a screenshot).
– Morin
learn how to read. k thx
It says it calculates how many pages to cache based on the amount of memory, not that it actually uses that much memory.
But Mozilla, supposedly the same code, doesn’t suffer from the same leaking. FireFox starts with 15MB on my system, 4 windows open, about 40 tabs. After 24 hours it’s bloated to 30MB and counting. Mozilla isn’t that bad at all.
Mozilla has page caching? Are you sure? I’m pretty sure it was only implemented in Firefox 1.5, but I may be mistaken.
4 windows, 40 tabs you say ? Maybe there is sth. terribly messed up (how? dunno – only 4 extensions, no themes, not even using it actually) with my 1.5 install, becouse my top raports 31 MB residual memory 1(!)window 1(!)tab (this very site) just after start-up. That is of course I didn’t perform any tweaks, but than again – why should I ?
-happy Opera user (no, not a troll)
The alternative is that he’s just trying to drive up traffic to get more ad revenue.
Said a few times already– OSNews doesn’t evolve around ad revenues like so many other newssites. OSNews is a completely voluntary endavour, we (Eugenia, Adam, Andrew, and me) get paid nothing for the work we do here. If our hosting was free, we’d get rid of ads altogether. In other words, all our revenues go to hosting, and that’s it.
And yes indeed, English isn’t my native language (Dutch is).
and chops the quote to deliberately make Firefox look bad
You guys are argueing silly semantics here. But hey, if you truly think I am on a quest to make Firefox look bad, then you are not even worth talking to. Hey look, a black helicopter!
And oh yeah, I have to add ‘anti-Firefox bias’ to the list of biases I supposedly have. Weird, how is it possible that ‘pro-Firefox bias’ is on that list too?
YOU ARE NOT EVEN WORTH TALKING TO!
YOU DON’T KNOW NOTHING!
HOW DARE YOU QUESTION ME IN ANY WAY!!
I CAN NOT CHANGE THE TITLE OF AN ARTICLE!
I AM WORKING MY ASS OFF HERE AND GET NOTHING, GODAMMIT!
I AM GOD!
I LIKE APPLE PIE!
I AM THE THOM!
Yeah, here we go again. I’m not pro- or anti- anything because I get accused of being on both sides!!!
Well, maybe you’re just pro-sensationalism, pro-misrepresentation, and pro-poor quality journalism then. And add anti-good English to that list, while you’re at it.
To clarify: I don’t think Thom is anti-Firefox at all. I think he is pro-flamewar inducing article titles. I suppose that is just good journalism, to drive up the amount of people who read it, but it gets annoying after a while when the titles are so obviously completely different than the articles. I happen to actually like what Thom says in most of his comments and have defended him in the past from bashers. But the titles he puts on some of his articles are just indefensible, and I wanted to give my opinion about how OSNews could be better in the future.
I think he is pro-flamewar inducing article titles.
Well, you’ll have to trust me on this, as there is no way I can disprove nonsense like the above: I’m not making headlines specifically to induce flamewars. If I really wanted that, I would’ve made the headline something like this:
Lead FireFox Developer Approves of Memory Leaks
Or whatever. This headline made the most sense to me, because it captures the essence of his blog post in as minimal an amount of words as possible. Because, I only have one line to summarize an entire article. That last thing is something many of you seem to forget.
Edited 2006-02-15 19:42
Firefox 1.5 page caching: not a “memory leak”
Try that on for size. Anyways, it’s not a memory leak, and you still don’t know what a memory leak is, so you shouldn’t go around saying that something is a memory leak when it isn’t.
Or maybe you just copied the title from slashdot, eh? Here comes the “I’m not responsible, I’m just quoting” excuse.
and you still don’t know what a memory leak is
Eh… How do you know I don’t know what a memory leak is? I don’t recall any studies saying that mind reading is possible, so I’m quite curious!
Firefox 1.5 page caching: not a “memory leak”
Yes… That’s a good title too. Could have used it… But I didn’t. If you weren’t so harsh from day one, accusing me of anti-Firefox baises and what not, I would have changed it, because this is indeed a better title. But, I don’t listen to people who start off with unsubstantiated accusations and insults.
Let this be a lesson for you, my friend: if you want something from people, treat them with respect and they’ll listen to you. Treat them with disrespect by throwing accusations and insults, and they won’t listen. That’s how it works in the real world.
In summation, a kind and normal email on your end would have yielded a 100% success rate. Insults and accusations made in a forum give you a 0% success rate (or, a 100% success rate if you just wanted to insult me in the first place, which is quite likely).
Or maybe you just copied the title from slashdot, eh? Here comes the “I’m not responsible, I’m just quoting” excuse.
No, I don’t read Slashdot.
I’m prepared to believe you just read your title incorrectly, as I said in my first post. But the one you selected is just wrong.
Read it carefully: Firefox Memory Leak a ‘Feature’. It is saying that a memory leak was purposefully added in as a feature. Which doesn’t make any sense, but that is what it says. Furthermore, you put ‘feature’ in quotes, which denotes sarcasm or doubt. At least when I read it. Clearly you are trying to stress that word for some reason since it is the only one quoted.
If you had instead put the quotes around ‘Memory Leak’ I would have had no problem with the title. A better title I think would have been: Firefox feature mistaken for memory leak
Clearly you are trying to stress that word for some reason since it is the only one quoted.
Quote marks can indicate two things: sarcasm, or, more likely, well, quoted text. In this case, I used them because of the latter.
OK, I am prepared to admit I overreacted a bit and this will be my last post on the subject.
Quote marks can indicate two things: sarcasm, or, more likely, well, quoted text.
Is it actually proper English to quote a single word from an article like that? If so, it’s certainly misleading. Consider the following two sentences.
VP Cheney accidentally shot a man yesterday.
VP Cheney ‘accidentally’ shot a man yesterday.
I’m sure in his press release Cheney said ‘accidentally’ somewhere, but if you quote that in the headline it seems very sarcastic.
“Bush vows to bring ‘Evil-Doers’ to justice”
This one doesn’t seem sarcastic, does it? Just emphasizing a term the object (Bush) used.
Edited 2006-02-15 23:55
It sounds sarcastic. Single quotes are commonly used for this type of denotation. Double quotes are more common to use for quoting a person. Even if it was double quoted, quoting one word is not only unnecessary but misleading.
I like your title better actually. Yours has some humour in it.
You added the quotes later on right? I’m not just losing my mind, the quotes weren’t there before?
I have to say, you make some crappy titles; but this isn’t one of them. I read it as saying:
“The thing people think is a memory leak is actually a feature.”
Which is pretty much what the article is explaining: How that’s even possible, and why it’s a feature they readily notice and probably like.
More biased, and harder to argue with, would have been this title: “Firefox Memory Leak”
I’m going to use an analogy I heard recently.
When dealing with murder in the judicial system, they punish based on intent. They don’t assume intent, but try to determine it first.
Yet, something as simple as interpreting a strangers intent in what they type, people on the internet tend to take the negative route. Instead of presuming good intentions, they assume bad. Unless you have very good reason to believe bad intentions, it’s kind of ridiculous.
Thom gains nothing from purposefully putting a “flaming headline” in a story. He doesn’t make more money. The site probably doesn’t gain anymore readers, and if they do, it isnt remotely significant. In fact, the only thing you get is a story with a lot more comments. This means more strain on the server.
If you read the headline, you can read it multiple ways. It could have been worded better, but why are we so concerned with that?
First, the headline can’t contain any special formatting for certain parts, so the ” was most likely put around ‘Feature’ too add emphasis.
Second, the summary right below quotes ‘memory leak(s)’, and further goes to explain that what is believed to be memory leaks, is in fact not.
Now if the summary left these things out, then I might agree that this headline was purely flammatory. But you have to read the headline and summary as a whole, at the very least. You should also read what is linked to get an even better understand of the story before making any judgements.
If you don’t, that’s your own damn problem, not Thom’s.
edit: note to ma_d: I agree with you and was just trying to expand on what you said
Edited 2006-02-15 23:06
Yea, I can tell you were just expanding. Thank you for explaining this better than I have/could!
J-S-H: Well, maybe you’re just pro-sensationalism, pro-misrepresentation, and pro-poor quality journalism then. And add anti-good English to that list, while you’re at it.
Man why are you so pissed? Come on dude learn to take it easy in life. Thom took the heading from slashdot so it is not really his fault. Also some time sensationalism is all that we want.
May be this will wake firefox developer to do something about their so-called cache which is a problem for many people. May be they need to dump the in-memory pages directly on disk in binary format so they can easily upload again without parsing etc. I don’t know what solution will be the best but their current solution is not good and they need to design something better.
Thom took the heading from slashdot so it is not really his fault.
Agreed with your post, but no, I did not take it from slashdot, since I don’t read Slashdot. If it’s the same headline, well, coincidence seemingly exists.
So firefox uses memory to keep frequently used information available for quick access. Seems quite reasonable. Seems like the reason I bought the ram.
This whole thing seems like it is caused by people using terms that they do not understand. “Memory Leak” and “Uses more memory than I would like” are two very different things. If people would just say what they mean instead of trying to sound smart by using terms that they do not understand this would be more productive for everyone.
If the original complaint was uses too much memory they could have got right to looking at ways to improve it. Instead we have false claims of memory leaks that lead to people arguing back and forth over what they are looking at.
That is not to say that there are no memory leaks, but unless you know how to track down what it is it is best to not to try tell programmers the cause of the problem. Stick to telling them the symptoms and let the people qualified to diagnose the causes do their job.
Edited 2006-02-15 18:51
Sorry, but it’s a leak.
The program has something to do, it malloc()s memory to do it, and after having done its job, free()s the memory “just in case some other program needs it to run”.
Firefox only grows, never shrinks. Try opening a lot of tabs (~100) and closing all but one…
And no, that’s not the cache. Try squid, to understand how a cache should be behaving, you’ll see it grows up to a user-specified memory usage and stays there.
That’s not firefox-specific, mozilla had the same problem 3 years ago, so the cause might be deeply rooted in the XUL or Gecko code.
I have observed carefully firefox 1.5.1 and earlier versions by means of “Windows Task Manager” and “Linux System Guard” after modifying both to show peak memory usage of a thread like firefox.exe (firefox.sh in linux); and found a huge amount of memory used and low delta between both.
eg: 17 tabs of firefox 1.5.1 on windows xp > 191 MB peak, now close all tabs but one and check the memory used it will not be 20 MB but 150 MB.
On linux the situation is even worse with the same web sites tabs opened it will consume 220 MB.
I guess more tweaks and code optimization is needed to get it more competitive before IE 7 hits the market with its smooth performance.
Do this for yourself: On windows
1. Run: taskmgr.exe
2. Go to the menubar and choose “view>Select Columbs” and check “memory usage” and “peak memory usage”
3. Now open firefox with only one tab (session saver users must close all tabs but one and restart firefox)
4. check both memories (write them on paper)
5. open more tabs (say 12 with your same thread)
6. Check both memories
7. Now close all tabs but one
8. Check memories and notice the inefficiency.
On Linux Run: “sysguard” and do the same as above.
Edited 2006-02-15 20:50
…of the problem, it looks like the issue is mislabeling. The retention of previoud pages in memory (to save re-rendering) does make sense as a way to improve performance. It would naturally lead people to presume that it is a memory leak when it is not. It would, however, be a memory leak if the pages still remained in memory longer than intended.
This aside from any other performance issues.
LOL!
I love to beat a dead horse but this is worse than my constant Linspire bashing…
go find something else to whine about!
If opera is better then use it!
Do a memory test or something!
I use firefox, dont use java, dont use any of the media plugins, no extensions or anything like that and I do not have any problems with memory usage. I close my browser maybe two to four times a day. The systems I use it on vary from a 450mhz with 256mb of ram to a 3.06ghz with 752mb ram. I usually have about 10-20 tabs open and rarely see more than 64mb of ram usage from firefox. When I close tabs SOME of that is released each time down to about a consistant 48mb after browsing and closing tabs. Yes that is a bit more than the 20mb or so that it starts up with but still not bad at all.
Would you like screenshots with that? Have you tried a clean install and verified the problem is FIREFOX and not plugins, extensions, yoru system, something else?
I clearly remember typing on a C64 in 1987, using a 300 BAUD modem to get text (no pictures) to scroll across my 80 column screen and chat with people a half-world away in a way that was previously impossible. It was just this side of magic. Well, it was when it didn’t crash or lock up or just get kicked by Ma Bell. It was also 9 bucks a month for 22 hours access time, a dime a minute after that. I may not like it if FireFox locks up (which for me, it doesn’t, go figure), but I’m REALLY hesitant to hold a stopwatch up to it and then fire up Opera to repeat the same. Sure the “feature” may need “improvement”, but the source is open, ya geniuses. So, fix it. As my dear departed Dad would have said, “I wish that’s all I had to worry about.”
If you’ve not browsing particularly heavily and you’re using a high end PC then there’s not going to be a big difference in speed between Firefox and Opera. Use an older computer or open a lot of pages and the difference becomes very clear.
Opera 8.5 is generally very fast and responsive on my old 400Mhz Celeron with 256Mb, even when used heavily with multiple windows and pages. On the same system Firefox and IE slow to a crawl with just a few pages open and are very frustrating to use.
Even on a newer Athlon XP with 1Gb RAM the speed advantage of Opera becomes clear when using it heavily. I can use the links panel in Opera to open dozens of pages in the background simultaneously and it still runs fine with 100+ pages open in several windows. I find it very rare for Opera to become unresponsive, even when I’m trying to push it to it’s limits. In my experience Firefox tends to becomes unresponsive and a ridiculous memory hog long before that, if it doesn’t simply crash under the strain.
As far as I’m concerned, it stops being a “feature” and becomes a memory leak when I close all the tabs and the memory usage stays the same.
open firefox on a single webpage (/., osnews, anything).
check mem usage => ~20MB
open ~100 tabs
check mem usage => ~250-350MB
close the ~100 tabs
check mem usage, nothing has changed.
I do not think caching is the issue here, it is disabled, and the behaviour stays the same.
So, yes, that’s a leak.
Why are people voting this up? Seriously guys? Does no one here understand how an OS allocates and truncates memory?
Just because memory is free’d does not mean it can be reclaimed.
Just because memory is free’d does not mean it can be reclaimed.
Just because memory is free’d does not mean it can be reclaimed.
Just because memory is free’d does not mean it can be reclaimed.
Get it? If your program has received a new segment of memory, and you allocate it all, then free up most, but not all of it, you still need that block. And while the OS _might_ be able to resize it, it won’t necessarily do it immediately.
You cannot get an accurate portrayal of a programs real resource needs by the actual memory usage.
A memory leak occurs when a program allocates memory from the heap, then loses all reference to it without freeing it. If it free’s it, and the OS does not reclaim anything from that, the memory is not leaked: And yes, this is a typical occurance.
Trust me, if there were a major leak in firefox you’d know it. People wouldn’t be able to browse for an hour no matter how much RAM they had (you can’t buy that much RAM). If it leaked say 5MB on every page view and you had 1GB of memory you’d be able to look at less than 200 pages before you had to close Firefox.
It is probable that Firefox has a few small, low occurance, memory leaks. It’s not all that uncommon from a large program. But I really doubt it leaks at all on each page view. This is the sort of thing the firefox dev’s would be checking very carefully before each release!
Believe it or not but there are tools out there which can tell you when you have a memory leak. This isn’t some magic code review process…
No.
Firefox requires around 20MB of RAM right after startup, single url / tab opened.
Displaying 100+ _more_ urls in tabs requires between 200 and 400MB (maybe more, depending on what you open). Say 300MB.
Close all the tabs save the first one. I’m not saying you should be down to 20MB. I’m saying there _should_ be a noticeable decrease, perhaps making firefox’ memory usage down in the 100MB range. At this point, a 3MB decrease (what is reproductible here) does NOT count. Your argument is valid, however I cannot believe that Firefox free()s everything needed and nothing can be reclaimed at the end by the OS. There is a huge difference between 20MB[app]+64MB[cache] and 300MB, and other applications are able to shrink to actual memory usage.
Call it a leak. Call it a badly designed layer on top of malloc() and free(). Call it anything you like, at the end of the day, firefox never, ever shrinks like it should do, like other programs do. And that is exactly what people are complaining about : a web application slowing any current computer down to a crawl after a few hours’ use. The only solution is restarting the application (reminds me of something). And since other applications are able to do it _right_, I do not see why it would NOT be a friggin’ bug.
Firefox IMHO has pretty much always been a decent web browser. That said I’ve always found it quite sluggish to load, especially prior to the 1.5 release.
It is slowly catching Opera up in terms of it’s feature set but not in terms of it’s efficiency.
so much negative about FF. I have to say that I really like it a lot. I’ve developed a couple of extensions for and really like it (using 1.5 extension dev techniques, big improvement over old extension dev).
FF may not be perfect, but if you look at just how flexible the design is – all the great extensions and cool ideas that can merge right in (like Greasemonkey scripts). FF is, in my mind, more of an application platform right up there with Java. I know it seems different – and certainly things like JOGL aren’t there yeet – but then again, there isn’t a native HTML control in Java that is anywhere near as good as <browser /> tag in XUL.
So for all of you complaining out there, try to consider that FF is much more than just a way to render a webpage. It is an application framework with a hefty API and lots of great UI features and many flexible ways of “expressing” itself.
Certainly I do find myself cursing FF from time to time – and it does seem to have issues running on many Macs – but I really think FF is something worth supporting. I greatly appreciate the work by Mozilla!
AS IF it’s only the last 8, or 8 per tab, what have you – it’s just not freeing the memory and consuming more… (and there’s NO way the last 8 pages viewed should take 500 {censored} megs) and that goes for every gecko based browser the past two years regardless of which ‘flavor of the month’ it is. I have yet to find one that is actually STABLE and doesn’t “Leak” (or whatever they want to call it now) like a steel sieve. Multiple machines, multiple OS – usually in under ten minutes use, 20 minutes on the outside.
There HAS to be more to it than this new excuse, and that’s all it is, YET ANOTHER (TM) excuse. If you let it ‘run rampant’ you see zero disk access, 99%+ CPU use well before you even reach HALF the available physical memory. THAT’s a big flag something else is wrong…
The problem is excasterbated by actually saving files… and further aggrivated thanks to image saves being routed through the (buggy, bloated, hangs the damned browser waiting on file connect, shows what a crappy tasking model they use) download manager. (the most SPECIAL feature of most gecko builds these days… in the same way some olympics are special)
Which is why I’m running Opera Beta 9 preview 2 as my primary browser across the board – since Opera beta’s have been more reliable than any Firefox ‘stable’ I’ve tried. I’d have given up long ago on even having the steaming pile known as firefox/mozilla installed in the latest versions if I didn’t HAVE to for site compatability testing (and with the websites I maintain I’m quickly reaching the point of wanting to tell gecko users to go {censored} themselves)
… and after my encounter with bugzilla – specifically my entries on this very subject from over a year ago – I actively campaign against gecko as being little more than a fanboy obsession. Not only were there year old existing entries for Mozilla on this (making the problem older than firefox) they were all “unconfirmed”, and when I posted explicit instructions on replicating the problem the best response they could come up with was to take me to task for using the term ‘crashed’ instead of ‘hung’ – a distinction I’d not heard in 30+ years of computing.
At which point, {censored} bugzilla, {censored} firefox, and bend everything gecko related over the table and {censored} them too.
… and before anyone gets their panties in a twist over this position being ‘confrontational’ or ‘trolling’ – get over it. From where I’m sitting this problem has been around for 2+ years with no real progress towards a solution (unless you count them actually promoting the workaround of installing a ‘save state’ plugin so you can kill the browser, then resume where you left off)… at which point the only real option is to spread the word to NOT use it until they actually FIX it (God forbid) or it fades into obscurity.
This bug, feature, or whatever you want to call it has EVERYTHING to do with how Gecko handles caching of images in memory. This is something that other browsers are NOT doing to the extent that Gecko is, which is why these issues are not apparent in competing browsers. Images are stored UNCOMPRESSED in memory, and an 8MB 1280×1024 compressed JPEG image can EASILY uncompress to 40 to 80MB in system RAM or even BEYOND. Compound this with that fact that multiple images of this size *could* be on any given page, and that we are >intentionally< maintaining a cache of these uncompressed images in memory, and that NOT all memory is automagically freed once allocated to a program, regardless of whether it has been explicitly freed within the program or not – and here you have the problem.
Before you go running your mouth next time, why not attempt changing your gecko based browser customizations to DISABLE memory-based caching:
browser.cache.memory.enable = FALSE
And you will see that memory consumption is on par with other browsers, including your beloved Opera.
I agree that enabling this feature by default, and not making a clear method of disabling this feature within the engine can have some semi-disastrous effects under certain (uncommon) operating conditions.
Or, stop wanking to 1600×1200 digicam shots all day and maybe you won’t have a problem.
>> Before you go running your mouth next time, why not attempt changing your gecko based browser customizations to DISABLE memory-based caching:
browser.cache.memory.enable = FALSE
Which has ZERO impact on the problem at hand – AND makes gecko based browsers run like CRAP speed wise. (If I wanted to browse the web at 386 speeds, I’d be using a 386)
That does NOT fix the problem in my experience, and it’s YET ANOTHER(TM) bullshit workaround that DOESN’T WORK.
Of course I love how people who cannot actually attack the content of the post resort to attacking the poster… when did we get to slashdot?
i dont recall ever seeing firefox go above 100MB ram. but then i dont have more then about 380MB total
right now its hovering around 70MB tho, but i have seen other programs swallow up more.
and this is with 12 tabs open right now.
changing said setting didnt do squat for me tho, probably because i allready have so low a ram amount…
If memory leaks are a feature, I wonder when bugs would become an option.
Of course it’s a feature. That’s why other people have to write ‘memory leak detection tools’ for the Mozilla team, because they obviously don’t know how to use malloc().
http://www.squarefree.com/2006/01/13/memory-leak-detection-tool/