Firefox 1.0 has major memory leaks

[vcv]

Firefox loads pages into memory. When Firefox closes, most of it's memory allocations are deallocated.

Really? Wow, I never knew!

Extensions CAN allocate their own memory,

Really? Could you point me to some references to show how an extension does so?

and if badly coded, may not close them. Firefox cannot be held accountable for all it's third party extensions.

It should be held somewhat accountable. It shouldn't let it happen in the first place I don't think. Even so, you can probably see that I got the leaks with no extensions installed.

Plus nowadays we a thing called Virtual Memory. This makes memory leaks much MUCH less of a problem. Memory leaks aren't *real* memory leaks until all your virtual memory is depleted. How often is this a problem? Rarely. With Firefox? Never.

A memory leak is a memory leak, whether in Virtual or Physical memory. It is a problem because the more the program is used, the more memory leaks, the more performance degradation.

You originally stated that this is not an attempt to troll, rather an attempt to warn others about Firefox's major memory leaks.

According to your own tests and results:

Firefox does not have "thousands" of memory leaks like you originally stated. In fact, Firefox has less than 10 memory leaks. It is Firefox's third party extensions that have thousands of memory leaks.

Yet you still claim you are right and that Firefox has major memory leak problems. After your own tests revealed Firefox does not have major memory leak problems, you still contest that it does. This sounds more like trolling than a warning to me.

Wrong. I got the memory leaks with no extensions installed as well.

[bgeh]

Not Found

The requested URL /pub/mozilla.org/firefox/nightly/2005-02-22-07-trunk/ was not found on this server.

585516244[/snapback]

well it works for me here :)

[NienorGT]

ROFL...

Nothing is perfect... they will correct it down...

[AxelStone]

Oye, out of curiosity I checked back. I think this thread should be closed due to it serves no more purpose, unless you consider arguing back and forth about thousands of leaks from obscure FF extentions. (Not arguing there probably are "leaks").

[vcv]

Oye, out of curiosity I checked back. I think this thread should be closed due to it serves no more purpose, unless you consider arguing back and forth about thousands of leaks from obscure FF extentions. (Not arguing there probably are "leaks").

585517318[/snapback]

How many times do I have to repeat this:

I got the leaks with no extensions installed.

[Pink Floyd Veteran]

I agree with supernova to do the test with clean profile and that is insane to have all these mb not freed even with 20 ext installed

[RyanVM]

vcv - file a bug in Bugzilla and include your results you posted recently.

[Jimny]

vcv, it is very apparent that you really don't know what you're talking about. You ARE trolling and everybody hates you for your ignorance.

[bkellner]

Really?  Could you point me to some references to show how an extension does so?

It should be held somewhat accountable.  It shouldn't let it happen in the first place I don't think.  Even so, you can probably see that I got the leaks with no extensions installed.

A memory leak is a memory leak, whether in Virtual or Physical memory.  It is a problem because the more the program is used, the more memory leaks, the more performance degradation.

Wrong.  I got the memory leaks with no extensions installed as well.

585517061[/snapback]

1. Most extensions have included .jar files. In these files is Java, which I shouldn't have to tell you has the ability to allocate memory. Everytime it makes a variable, everytime it creates a string, it allocates it to memory.

2. If you hold Firefox accountable for not deallocating memory that it's third party extensions allocate, why not hold Windows accountable for not deallocating memory that third party programs like Firefox leaves? In that case, why not say the programming languages are at fault for not automatically deallocating memory when the program is done running?

I highly doubt you will have to worry about a few memory addresses not being deallocated. But to say Firefox has THOUSANDS of memory leaks is a lie. It is the extensions which has THOUSANDS of memory leaks. That's why when you uninstalled all the extensions you only have 7 or 8 heap leaks. Which is barely, if at all, a problem. This means you would have to open Firefox, browse websites, and close it over 200 times before you reach even 2000 heap leaks. That is still a long way away from the leaks you got which the extensions. Thus, this problem is VERY minute, and solely the extensions. Firefox CAN NOT and SHOULD NOT be held accountable for what the extensions do or do not do. If the extension creators are sloppy, then, well, there you go.

Major memory leaks? I think not. We don't run on 64KB memory anymore.

[vcv]

In these files is Java, which I shouldn't have to tell you has the ability to allocate memory.

The extension is not managing it's own memory. Java has GC. Though, I was under the impression FF uses javascript, not java.

Everytime it makes a variable, everytime it creates a string, it allocates it to memory.

This is different then allocating your own memory.

AGAIN, for the 3rd time AT LEAST.. I got leaks WITHOUT ANY EXTENSIONS.

edit: Firefox does indeed use JavaScript for it's extensions, not Java. Regardless, JavaScript code is not capable of manually allocating memory.

[takkun]

vcv - file a bug in Bugzilla and include your results you posted recently.

585517576[/snapback]

Yes, that's what you should do, at least there, something can be done about it, with your research, and i think most FF users know that it has random memory leaks, [not for me, i've only seen that 2 or 3 times i have a cpu/ram monitor on my desktop, i use FF all the time, my regular uptimes are 3-4 days and even if FF is leaking memory my performance is still perfect, so i don't see how this can be a real problem] but those who don't could use the info.

[Xtreme $niper]

I've never had this problem. My firefox never has gone over 35mb of ram even with more than 30 tabs open at the same time with 3 extensions. And when I close firefox, the ram is no longer being used. Simple as that.

[samriley]

just out of interest what extensions were you using, im always amazed when people say they have to use 10+ extensions with fx. personally i only use 2, adblock and web developer (makes webdesigning easier) so if i wasnt a webdesigner i wouldnt use that either

[vcv]

vcv, it is very apparent that you really don't know what you're talking about.  You ARE trolling and everybody hates you for your ignorance.

585517672[/snapback]

Yet you can't prove me wrong, and have failed miserably trying, instead making yourself look foolish.

vcv - file a bug in Bugzilla and include your results you posted recently.

I will once I get more information on why and how this happens.

[bilemke]

:o OMG!!!!! 24MB in a leak..

Wow, why did they give this the "1.0" final name/number? Anything that leaks 24 mb of memory after that little use has got some issues alright. I must test this on my system. I have a gig of ram but I pride myself on not having to use junkie little programs to recover memory and not having to restart that often. This is not acceptable. :no: I have tryied the nightly builds and will test them for this too if I get time but I have not like the little problems I have had witht he nightly builds so have always gone back to the 1.0 final.

That, and the number of times I have it crash on me on a regular basis with more then 10 tabs open..

This might sway me to put Firefox on the back burner and swith to using Opera and IE together more. VS now, I am currently using all 3 almost equally.

With all the pre 1.0 builds that were made and tested, this is very disheartening.

Edited February 22, 2005 by bilemke

[Schmoove]

Yes I have it too. I often close Firefox and find it eating up 140mb or more.

You can say that it is the fault of an extension, but still Firefox should manage that better.

.... and YES a browser that uses 140mb when it has been closed down IS a big deal. It is just a stupid bug they need to iron out.

[vcv]

:o  OMG!!!!! 24MB in a leak.. 

Wow, why did they give this the "1.0" final name/number? Anything that leaks 24 mb of memory after that little use has got some issues alright. I must test this on my system. I have a gig of ram but I pride myself on not having to use junkie little programs to recover memory and not having to restart that often. This is not acceptable.  :no:  I have tryied the nightly builds and will test them for this too if I get time but I have not like the little problems I have had witht he nightly builds so have always gone back to the 1.0 final.

That, and the number of times I have it crash on me on a regular basis with more then 10 tabs open..

This might sway me to put Firefox on the back burner and swith to using Opera and IE together more. VS now, I am currently using all 3 almost equally.

With all the pre 1.0 builds that were made and tested, this is very disheartening.

585517811[/snapback]

I wouldn't ditch FF over it quite yet. Windows does a good job of cleaning up still, and unless you notice performance issues, I wouldn't worry too much about it.

[Ciderx]

Yeah, everyone knows there is are bad memory leaks in Firefox (there are also some 40000 returns for firefox memory leak on Google), and to be honest, in the hype and ego of the latter stages of development, there was no interest in fixing this.

But, y'know, if free, open source, not allowed to complain about it, etc etc etc

[Jimny]

you can't prove me wrong

It's your job to prove yourself right, which you have not done. Ranting and raving about the log you posted etc does not prove yourself right.

Just because memory was allocated 2000 times and none of it was deallocated does not mean there are 2000 memory leaks. It could just mean that the same code that allocated it was called multiple times.

I will once I get more information on why and how this happens.

If you have to figure out how a memory leak happens...you really shouldn't be starting threads discussing them.

[vcv]

It's your job to prove yourself right, which you have not done. Ranting and raving about the log you posted etc does not prove yourself right.

I posted a log from a memory leak program, I posted a screen-shot, I posted a link to the program I used AND I posted instructions to reproduce what I got. WHAT MORE DO YOU EXPECT?

Just because memory was allocated 2000 times and none of it was deallocated does not mean there are 2000 memory leaks.

Uh.. yes it does. This program keeps track of what data segments were allocated and not freed. It does not have access to the program source, so it cant simple look at code like you suggest.

If you have to figure out how a memory leak happens...you really shouldn't be starting threads discussing them.

:blink: Now I'm starting to think you're just trying to be obtuse.

[neoufo51]

Yeah, thats why I download the nightlies every couple of weeks or so.

[phend-one]

That's a problem I see with the testing. Nightly builds. Make sure if you're doing this testing, you define what scope of testing and issues you wish to address. Basis for comparison are useful if, for example, they are based on previously released version with the current version. Nightly builds change constantly and do not provide an accurate measurement for benchmarking.

The average user doesn't download nightly builds everyday or often. The average user pays attention to regular releases/updates, if even. Then again, the average user doesn't even know about memory leaks (just look around on the posts about people checking task manager about memory usage and not about deallocated memory). Rest assured, when something does go wrong, people will be blaming "Windows." Who gets the bad reputation now? Surely not the coders of the program causing the memory leaks. Everything is blamed on "Windows."

That's why good programming is important. End-users expect things to work flawlessly. They don't care about how things are coded; they're not very educated about these issues. It's up to the programmer to write good code.

That being said, whatever your user experience is with Firefox, great. If you think it's not up to your standards, you don't have to use it or you can wait for a new version to try again. You could also try alternative browsers.

As some users are stating: the purpose of this post is pointless. Is it really? I think it's been informative. Has the author proved his point? He/she has some good points. More constructive critisism is required, not flaming. Nothing gets solved that way.

To the author: some great minds had to face great critisism before their views became accepted. Don't let the comments get to you. Do what you set out to do, provide an objective analysis of the issue, and then present it to the public. If they decide to accept it, that's up to them. You did your part, which is more than a lot of posters have done in this thread.

[Jimny]

Sorry VCV but your trolling attitude and pointless threads aren't making people switch from firefox to opera like you wanted.

[vcv]

Sorry VCV but your trolling attitude and pointless threads aren't making people switch from firefox to opera like you wanted.

585518508[/snapback]

I could care less which browser a person uses, I use both myself. It's a shame that in this very thread I recommend to someone that they NOT stop using FF when they said they were going to, isn't it?

You're making a fool of yourself, please stop now.

----

Thanks for the words phend-one. I am attempting to do what I can to narrow down this issue. I tried for a short bit earlier to download the latest source and compile it on my windows boxen, but it was spitting out errors (says I need w32api 2.4 or higher when I have 3.1, among some other errors), so no luck on that today. I will keep trying until I get it working though.

Once I have my sandbox set up, I will compile with debug info, and maybe work some custom memory monitoring into the code, and go from there. In my own personal project, we have a memory profiler that integrates right into the code in debug builds by hooking calls to new/delete and malloc/free, then reports any memory leaks when the program is finished. I hope to put something similar into firefox so that it can tell me EXACTLY where this is occuring.. something that would be extremely useful for the FF devs.

Edited February 23, 2005 by vcv

[redfish]

Give it some time before you complain about the memory leak.

585515088[/snapback]

ok..

Give it some time before you say Firefox is superior to IE in every way.

(maybe not you but other people)