@wawa

Some bugs just so worse, that by plain "memory handling" tool you can't catch them.

Not the latest moment, that we not have at all on any OS (3/4/mos/aros) any _REAL_ tools for debugging. On os4 we have initial attempts (db101), but its at the begining of way. Also someone tryes to works to make the same as "valgrind", but its all "someone start maybe when 2 more weeks" :)

With something like valgrind , for sure works will be easy and faster.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@wawa

And do you think that they are magically resolvible just porting OWB MUI or OWB Reaction to OS3 ?

I don't think so, as OS4 is mutch more robust and modern that any OS3/AROS out there

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@kas1e

Quote:

Not the latest moment, that we not have at all on any OS (3/4/mos/aros) any _REAL_ tools for debugging.

Uhm, well... for OS3.x there are some dozen very good debugging tools Several source code debuggers (gcc is not compatible to them, but this is definetly gcc´s fault), disassemblers, recompilers, memory tracking tools, the invaluable Enforcer, the MU tools and a lot more even I dont know.

I really hope that db101 will evolve over time, as this is the most promising way to get a useable source code debugger that supports more than gcc generated code. Uh, and that would be much more user friendly than gdb could ever be

@whose
Quote:

Uhm, well... for OS3.x there are some dozen very good debugging tools Several source code debuggers (gcc is not compatible to them, but this is definetly gcc´s fault), disassemblers, recompilers, memory tracking tools, the invaluable Enforcer, the MU tools and a lot more even I dont know.

Disassemblers/recompilers/hexeditor - that is not problem. They all there and so on. Did you see Valginrd on linux ? That what i mean. Os3 enforcers can just suck something in compare (but of course, better than nothing sometime).

Anyway, all that porting bugs are fixable, we will keep you all informed.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

I dont think porting OWB to OS3 or AROS would help because OWB on MorphOS gives good reference. I mean, by comparing those three it should be fairly easy to figure out if it is the OS core leaking memory, or is it in Webkit or just the GUI.

_________________
Amiga Developer
Amiga 500, Efika, Mac Mini and PowerBook

@itix
first of all im not a programmer, so everybodys word here outweights mine by magnitudes. my "simple mind" tells me that though even if mos is a good reference (since i take most bugs are not present there) further implementation odf (mostly) the same codebase could help to debug it better. feel free to ignore my post or teach me the better. i do not try to dazzle my agenda, to get (fast, as it seems ) owb on os3.
;)

Last edited by wawa on 17-Feb-2011 at 09:55 AM.

guys

I've been using morphOS OWB for some time now and never seen memory leaks like you guys are referring to.. ???

@realize

I can confirm they are there under Morphos version as well.
So it looks like a webkit issue.

_________________
If your nose runs and your feet smell, you're upside down.

@TiredofLife

It's no secret WebKit leaks memory over time (on all platforms, and it's not the only one, FireFox also leaks like mad). But i've never seen a case where staying idle on a site depleted memory, though.

But under MorphOS, all memory is returned after quitting program, though.

@Fab

Yeah, I have asked for some examples of memory depleting when staying idle on a site but nobody has managed to provide me with one I can easily test.
Wunderground used to do it but doesn't seem to do it anymore.

Both AOS4 and MOS versions, return the memory after closing down the browser though so it doesnt appear to be a big problem.

I know this isn't the case for some users, so they may need to continue investigating.

_________________
If your nose runs and your feet smell, you're upside down.

@TiredofLife

Well I think people made WAY to big of a deal with this memory leak. Especially this one silly site. I mean, why Hijack a very good and productive thread and talk about this so much?

@TiredofLife

Quote:

TiredofLife wrote: @Fab Yeah, I have asked for some examples of memory depleting when staying idle on a site but nobody has managed to provide me with one I can easily test. Wunderground used to do it but doesn't seem to do it anymore. Both AOS4 and MOS versions, return the memory after closing down the browser though so it doesnt appear to be a big problem. I know this isn't the case for some users, so they may need to continue investigating.

You need a website that automatically refresh itself now and then i believe. All i know is that i have sometimes forgot OWB running with some website tabs open only to find memory nearly or completely eaten up by next morning.
I am not sure which exact website(s) were the culprit those times though.

@realize

Quote:

realize wrote: @TiredofLife Well I think people made WAY to big of a deal with this memory leak. Especially this one silly site. I mean, why Hijack a very good and productive thread and talk about this so much?

Probably because people in this very same thread asked for more info so they can figure out what exactly causes this leak, as that is still apparently very much unknown.

And i believe it is kind of a serious problem when it leads to you having to reboot whole OS in the end because of lack of memory or because system froze up due to memory+swap was eaten up.

@realize

The memory leak for me isn't the big issue.
The serious problem is for a few users who don't have the memory returned.

If it's linked to OWB 3.31, the hope for some is that the version from Kas1e/Fab/Denil/Slash will not have this bug.

From my observations though, it's not OWB that is the problem when it comes to memory not being returned.
Although I wouldn't rule out a combination of OWB and the way some people have their systems set up.

But as I said, those people will need to do further testing.
Or they can wait for the new version and see if that helps.

I agree though this thread shouldn't have been hijacked.
I have been guilty in joining in with that and for that, I apologise.

_________________
If your nose runs and your feet smell, you're upside down.

To get this thread back on-topic, it's probably worth reminding ourselves of kas1e's latest progress report from 5th February:
Quote:

@all mui-owb progress: Deniil found how to replace some necessary mui4 methods on mui3, and now check this out: google maps mui-owb & reaction-owb at one time some "heavy" site And video of current port here (youtube as usuall) Still need to fix stuff (you can see DSI on video), also there is some more DSIs there and there , some problems with history of urls , tabs for now a bit ugly, but well, text typing not works normally (need to rewrite some encoding fucntions to os4), and threading still not fixed, which can give us random dsi , favicons and fast link also need to fix and so on. In others words, while it looks like "its almost done", there is many to do.

_________________
Author of the PortablE programming language.
It is pitch black. You are likely to be eaten by a grue...

@all
Some progress about mui-owb:

- Mason almost done all the graphics work for port (we need to use new images for). And there is 2 screens with his images (toolbar, status bar, progress bar, url bar, etc):

screen1
screen2

- Deniil yesterday fix some some stuff, which solve for us loading of last commented child - fastlinks. But fastlinks based on bookmarks, so while we not fix bookmarks normally, fastlinks also not works visually (but technically they are now).

- Slash still working on Process class , and almost done with

- Tboekel help us as well with fix some bugs and point us on some new ideas about some classes which we can use

ps. that post from mui-owb (but while i post i fear it will crashes :) , so still port-bugs still here, and they all should be fixed before release.

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

Quote:

Related to speed, its 100% not slower than reaction one, and maybe even a bit faster. But that can be placebo, so if not faster, then speed for sure the same (if not a bit faster still) (i am about speed of rendering).

To be a bit more precise, OWB MorphOS can only be faster, for design reasons:
- Refresh method uses some heuristic to reduce the surface to render when possible and when it's profitable to do so.
- For pages with fixed elements like facebook status bar, scrolling should really be substancially faster than OWB Reaction, because it scrolls using blitter first and then redraws the fixed elements later, instead of redrawing the whole page, like OWB Reaction would do (what i'm talking about can't apply to pages with fixed backgrounds, though).

@kas1e

The new splash screen is awesome, thanks boyz

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6

@Fab

Can you please give one more time 2-3 urls which i can test now to see redrawing differences (you give me one in irc few days ago, but i just loose that logs :( ).

_________________
Join us to improve dopus5!
zerohero's mirror of os4/os3 crosscompiler suites

@kas1e

Try opening a few tab here: http://amiga.ikirsector.it/forum/index.php

_________________
BACK FOR THE FUTURE

http://www.betatesting.it/backforthefuture

Sam440ep Flex 800 Mhz 1 GB Ram + AmigaOS 4.1 Update 6
AmigaOne XE G3 800 Mhz - 640 MB Ram - Radeon 9200 SE + AmigaOS 4.1 Update 6