Firefox 1.0 has major memory leaks

[Jimny]

Google: firefox xpcom extensions compiled

http://kb.mozillazine.org/Dev_:_Extensions

it is also possible for extensions to include compiled code in the form of XPCOM components

I know the JS can use XPCOM components, but can it use ones not included with mozilla or firefox?

As long as it's included with the extension, yes. Edited February 24, 2005 by Jimny

[vcv]

Google: firefox xpcom extensions compiled

http://kb.mozillazine.org/Dev_:_Extensions

585524419[/snapback]

Ok. Now can you write your own XPCOM component and have your extension use that XPCOM component by including it in the XPI?

If so, what extensions have actually done this?

XPCOM/XPConnect. Mozilla packages together useful functions into XPCOM (Cross-Platform Component Object Model) components, which may then be accessed from external code, including Javascript via XPConnect. This allows the extension developers access the preferences database, the filesystem, and many other pieces of Mozilla technology.

It's my impression that XPCOM is the way for extensions to access the different aspects of the core mozilla code (like it says above).

[Silverado Nomads]

Read through all the posts here.....I saw somebody post about windows emptying the memory the exe held once it's closed.

Windows doesn't do that by default really. It keeps all the dlls the exe calls up into memory for a short time before closing them.

Now there is an option to empty it out soon as you close your program(in the registry, but it's been a long time since i've used that option). But so many core windows components would crash because they wouldn't have access to there DLL that was there upon startup.

That may be part of the problem(i'm not saying firefox isn't the problem, i'm just saying that you should look at everything first instead of screaming fire everytime you find something that isn't as you would expect). Take it as you will.

- Khar

[golazo]

good find, i don't use the tabs anyway. so im coo :ninja:

[Jimny]

Ok.  Now can you write your own XPCOM component and have your extension use that XPCOM component by including it in the XPI?

If so, what extensions have actually done this?

It's my impression that XPCOM is the way for extensions to access the different aspects of the core mozilla code (like it says above).

585524449[/snapback]

Correct, XPCOM is how extensions communicate with mozilla, however it can also be used to add functionality not possible with javascript.

I'm not sure what extensions have done this, you'd have to download and unpack then, and check.

[hdood]

Read through all the posts here.....I saw somebody post about windows emptying the memory the exe held once it's closed.

Windows doesn't do that by default really. It keeps all the dlls the exe calls up into memory for a short time before closing them.

Now there is an option to empty it out soon as you close your program(in the registry, but it's been a long time since i've used that option). But so many core windows components would crash because they wouldn't have access to there DLL that was there upon startup.

585524461[/snapback]

While true in a sense, it is not very relevant, and it really only applies to the core operating system DLLs that are kept in the DLL cache in virtual memory for increased performance and integrity (See KnownDLL)

That may be part of the problem(i'm not saying firefox isn't the problem, i'm just saying that you should look at everything first instead of screaming fire everytime you find something that isn't as you would expect). Take it as you will.

585524461[/snapback]

This entire thread is pretty redundant, as thousands of Mozilla developers all over the world are likely aware of most of the current issues with Firefox (what?! you mean I'm not the only one capable of running profiling software??!!).

It all comes down to priorities, not all bugs are important.

One thing is for sure though, bickering about it on a site like Neowin, especially in a very non-technical fashion, doesn't benefit anyone. Maybe if the thread was about finding and fixing the bugs, but it isn't. The only thing it might do is scare people out of using Firefox, for no good reason.

[Fred Derf Veteran]

This entire thread is pretty redundant, as thousands of Mozilla developers all over the world are likely aware of most of the current issues with Firefox (what?! you mean I'm not the only one capable of running profiling software??!!).

It all comes down to priorities, not all bugs are important.

One thing is for sure though, bickering about it on a site like Neowin, especially in a very non-technical fashion, doesn't benefit anyone. Maybe if the thread was about finding and fixing the bugs, but it isn't. The only thing it might do is scare people out of using Firefox, for no good reason.

585526105[/snapback]

Now that is something that I can agree with.

[vcv]

Memory leaks are not important? I pity the Mozilla Developers if that holds true for them. I seriously have no respect for any Mozilla Dev that holds this belief. But I don't know any specifically that do ATM.

If the moz devs are aware of this and have been since the release of 1.0, and still don't care enough to fix it (for 1.01 at least), I find that disgusting.

But I've already stated in the thread that this issue should NOT be a factor in anyones decision to use or not use FF (or mozilla).

[hdood]

There are currently 6620 open bugs for Firefox. Why don't you offer your services, and fix the ones you care about?

[vcv]

I understand there are a lot of important bugs. But memory leaks like this (ones that can cause thousands of leaks) should be near the top of the list. Memory leaks have always been a priority on projects I've worked on, and my friends/co-workers have worked on.

Regarding fixing it myself.. well.. I honestly would try. But there's a few problems.

1. Aside from the new few days that I have to relax, I'm usually pretty busy with doing coding for my job.

2. I also have a side-project I work on that keeps me busy

3. There is a lot you need to learn before doing work on mozilla code. Lots of guidelines to read and what not. I've read a bit before, so I have a general idea, but they're very strict on their code, and it would be hard for me to adjust quickly.

4. I have tried compiling FF in the past few days, but been unable to get it to fully work. I've made some progress, but not enough. If anyone could help me get past the errors I've been getting, I'd be more than happy to look at the code and see what I can do.

If I was able to get it to compile, I'd likely put in my own memory checking code (very quick/easy to do actually) and find out exactly where it's occuring.

[+M2Ys4U Subscriber¹]

I understand there are a lot of important bugs.  But memory leaks like this (ones that can cause thousands of leaks) should be near the top of the list.  Memory leaks have always been a priority on projects I've worked on, and my friends/co-workers have worked on.

Regarding fixing it myself.. well.. I honestly would try.  But there's a few problems.

1. Aside from the new few days that I have to relax, I'm usually pretty busy with doing coding for my job.

2. I also have a side-project I work on that keeps me busy

3. There is a lot you need to learn before doing work on mozilla code.  Lots of guidelines to read and what not.  I've read a bit before, so I have a general idea, but they're very strict on their code, and it would be hard for me to adjust quickly.

4. I have tried compiling FF in the past few days, but been unable to get it to fully work.  I've made some progress, but not enough.  If anyone could help me get past the errors I've been getting, I'd be more than happy to look at the code and see what I can do.

If I was able to get it to compile, I'd likely put in my own memory checking code (very quick/easy to do actually) and find out exactly where it's occuring.

585526485[/snapback]

then, with all due respect, either submit the bug to bugzilla or quit whining about it here in a sesationalistic thread, where nobody can do anything about it.

[Brandon Live Veteran]

A few things

-Memory leaks have nothing to do with memory "hanging around" after the program is closed. Windows (or any modern OS) will free up any space allocated by a process when it exits - even if it crashes or is forced closed.

-This has little or nothing to do with the size you see Firefox.exe "using" in Task Manager's "Mem usage" tab.

-Memory leaks ARE BAD, but no code is perfect. And no profiler is perfect either.

-No one believes mircleman

-There are significant bugs in Firefox 1.0's memory management, and they seem to involve tabbing (as well as extension management).

-This is no reason to stop using Firefox if you like it and don't have any problems. It isn't going to do any permenant damage to your computer, and if it does appear to be causing slowdown at any time, you can always close it.

-If you can reproduce this problem predictably, please submit a bug report with exactly what you did. I'm sure the developers would love to hear from you.

-It sounds like several memory management bugs are being fixed every day. If you like Firefox, I suggest upgrading to the next point release as soon as it is released.

[vcv]

then, with all due respect, either submit the bug to bugzilla or quit whining about it here in a sesationalistic thread, where nobody can do anything about it.

585526603[/snapback]

1. There is already bugs submitted regarding this.

2. I am not whining, nice try though

3. I have not exaggerated anywhere. There are thousands of leaks under certain scenarios, and a spot in a code that has a leak IS major when that code is called thousands of times.

-Memory leaks have nothing to do with memory "hanging around" after the program is closed.  Windows (or any modern OS) will free up any space allocated by a process when it exits - even if it crashes or is forced closed. 

Mostly correct. It cleans up the heap for you. However, I only meant mentioned that the memory is still there at close to illustrate that it wasn't being free'd by FF.

-Memory leaks ARE BAD, but no code is perfect.  And no profiler is perfect either.

Of course, but when thousands of memory leaks happen simply by closing a tab in a 1.0 release of a product, that is bad. And I emailed the company that makes the profiler with the info to see if it's possible that the profiler has a bug itself.

-No one believes mircleman

:laugh:

-This is no reason to stop using Firefox if you like it and don't have any problems.  It isn't going to do any permenant damage to your computer, and if it does appear to be causing slowdown at any time, you can always close it.

Exactly. I didn't mean to scare people away from FF at all. But I probably should have been more clear and maybe not had "WARNING!" in the thread.

[Joel]

1. There is already bugs submitted regarding this.

585527158[/snapback]

Great, so let them fix it or ignore it.

2. I am not whining, nice try though

585527158[/snapback]

It sounds like you are. I've read this whole thread, and you can't seem to keep to one testing scenario and give results based on whatever higher number of "leaks" you achieved.

3. I have not exaggerated anywhere.  There are thousands of leaks under certain scenarios, and a spot in a code that has a leak IS major when that code is called thousands of times.

585527158[/snapback]

Cars blow up under certain conditions also. Does that mean that they all don't work properly? You've already said that Firefox should be responsible for the extensions' memory problems, which made no sense.

Exactly.  I didn't mean to scare people away from FF at all.  But I probably should have been more clear and maybe not had "WARNING!" in the thread.

585527158[/snapback]

Title edited

[+M2Ys4U Subscriber¹]

3. I have not exaggerated anywhere.  There are thousands of leaks under certain scenarios, and a spot in a code that has a leak IS major when that code is called thousands of times.

585527158[/snapback]

I havn't said that you exaggerated anywhere, just of whining and making your thread title sensationalistic (i.e. the "WARNING!" and taking whatever results were highest at the time.)

Everything else I wanted to follow up on from your post Joel has already said.

[vcv]

It sounds like you are. I've read this whole thread, and you can't seem to keep to one testing scenario and give results based on whatever higher number of "leaks" you achieved.

Cars blow up under certain conditions also. Does that mean that they all don't work properly? You've already said that Firefox should be responsible for the extensions' memory problems, which made no sense.

Title edited

585527241[/snapback]

Actually I explained exactly how to reproduce it. How is that not one testing scenario? And it'll happen for anyone that uses tabs, I think that the "certain conditions" here apply to a lot of people.

Also, yes, firefox is responsible for extensions memory (unless it uses its own XPCOM component, which I don't even know if it possible, and if so, most extensions do not do so).

Thank you for editing the thread title though.

I havn't said that you exaggerated anywhere, just of whining and making your thread title sensationalistic (i.e. the "WARNING!" and taking whatever results were highest at the time.)

exaggeration is part of the defintion of sensationalistic. And I didn't take the highest results actually. I ran the test many times and got the same number, give or take 50.

[Jimny]

Also, yes, firefox is responsible for extensions memory

Can you provide proof?

(unless it uses its own XPCOM component, which I don't even know if it possible

585527408[/snapback]

I told you that it was possible and provided documentation on how to do it. One extension that does is HTML Validator: https://addons.update.mozilla.org/extension...=Windows&id=249

[vcv]

Ok, thank you. So some do use XPCOM components. However, I know TBE doesn't and most extensions don't. Firefox is responsible for any memory in the JS files, because it is JavaScript.

So I was partly wrong. It is possible for an extension to cause memory leaks.

However, this was not the case in my tests.

You only linked me to a general XPCOM page, not any specific documentation on how to do it.