@AlexC

" I blacklisted IB ..."

Interesting.

IIRC, some MUI update broke IB awhile ago. Anyone have a link to a fix (or is it already fixed via Amiupdate, must check).

Other than that. I use IB because of it's speed. Must see if it's fast also without JIT.
(and I should study more netsurf for speedy browsing needs)

(I have 667Mhz SAM for AOS4 and Odyssey is not "fast" on it. Fully usable etc, but not something that I would call fast.)

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@KimmoK

No idea about recent MUI, but it's been stated for years that:

Quote:

The IBrowse crashes seem to be associated with the 68K emulation, as Fab suggested. We don't know the cause or which program is responsible. It might even be a MUI problem.

Source from 2011

Quote:

I shall repeat for the record that IBrowse never crashed, even during the period when everyone swore it crashed many many years ago. I used serial dialup. That's your clue.

Source same thread

Quote:

it just doesn't crash anymore with the native PPC port, but presumably the 68k 2.5 will be almost as crash happy as 2.4.

same source from IBrowse team member

Note that similar information has been repeated SINCE 2011 in these forums.
If any recent changes have affected this, I would not know.

#6

_________________
This posting, in its entirety, represents solely the perspective of the author.
*Secrecy has served us so well*

@Xmas87

From my own use of the two since getting my A1 working again a few days ago, IBrowse is faster. But Odyssey can render pretty much any page the way other browsers (Firefox, Chrome, etc) can, unlike iBrowse.

So for me it's either speed (IBrowse) or proper rendering (Odyssey).

but I don't consider IBrowse to be buggy in any sense that I've seen.

And back on topic, looking forward to multi-core support in OS4, even though I'm still on a single core A1-XE G4.

_________________
This posting, in it's entirety, is the opinion and/or statement of the author and does not reflect the views and/or position of this site.

@DWolfman

Pretty much what I expected. Thanks.

I too am looking forward to multicore support. However it would be much easier if there was some prospect or hint that it was definitely coming.
So far all I see is a vague 2010 announcement followed up with even more vague "it's coming" posts.

@AlexC

Thanks for this information. NutsAboutAmiga might be interested in this.

Just out of interest:
I understood you were using the 68k version with OS4, right?

What about the "limited OEM version of IBrowse that is included with AmigaOS 4"? How is this version regarding stability?

_________________
Ciao

Dandy
__________________________________________
If someone enjoys marching to military music, then I already despise him.
He got his brain accidently - the bone marrow in his back would have been sufficient for him!
(Albert Einstein)

@Samurai_Crow

Quote:

That's why PThreads makes fork obsolete (though still commonly used on Linux).

fork() uses some kind of copy-on-write mechansism and is part of the system. But unless PThreads is part of the system API and does the same thing or better I don't see it making fork() obsolete.

@number6

Since IBrowse works in MorphOS with MUI 4 crashes must be related to 68k emulation in OS4. If IBrowse authors have also stated that it is no longer crashing in native PPC version it is strong hint that there is something in the OS. It could be just very small mistake in instruction handling.

Maybe there is more. Last year I found out in Saku 2013 event that when I quit IBrowse in OS4 beta it reboots the machine immediately.

Last edited by itix on 14-Aug-2014 at 03:52 PM.

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

@Dandy

Quote:

Hmm - I use IBrowse 2.4 (on OS 3.9/WarpOS) since a decade or so and never noticed a bug.

That's most likely beacuse it has no memory protection and clobbered memory would go unnoticed. AmigaOS3 and the unbderlying hardware didn't reveal hidden bugs programs had.

On OS4 is is unaccetpable as it has memory protection. At least to an extent. So any bugs in IBrowse will show up. The other thing is that it is running on an emulator. The emulator may have a bug. But either way that doesn't mean IBrowse is innocent becaue it doesn't crash on an OS3 Amiga.

One would have to duplicate browsing with IBrowse on each system to see if it crashes on one machine and not the other. On OS4 it crashes a lot. Well my dad can make it crash while on a Google.

@Hypex

There is no more memory protection in OS4 than there is in OS3. In fact it is easier to spot illegal read/write access in OS3 because typically they have very little ram. Chances that random data is pointing to legal memory address is higher in OS4 systems.

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

@itix

Quote:

There is no more memory protection in OS4 than there is in OS3. In fact it is easier to spot illegal read/write access in OS3 because typically they have very little ram. Chances that random data is pointing to legal memory address is higher in OS4 systems.

There is no memory protection in OS3 unless you run special dev tools like enforcer, and enforcer doesn't offer all the types of memory protection that OS4 has, at the very least OS4 protects the executable memory area, and also read only strings and other constant program data in addition to the types of protection offered by enforcer.

Most people do not run enforcer day to day on their 3.x systems, not even in a WinUAE environment.

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

@Hypex

Quote:

fork() uses some kind of copy-on-write mechansism and is part of the system. But unless PThreads is part of the system API and does the same thing or better I don't see it making fork() obsolete.

The pthreads API (no capitalisation) is part of the POSIX (hence the 'p') API. It doesn't need to copy on write because the whole point is that is a *thread* and runs in the same data space as the parent.

I'm not sure if pthreads makes fork obsolete as such, but it is much better for many of the uses multi threading / processing is used for.

fork() is useful for doing different tasks simultaneously or spawn separate but unrelated coppies of the same task (server dealing with separate requests)

pthreads is useful for subdividing one task into separate parallel operations.


Take blender as an example, fork() is used to spawn a new process to play animations in parallel to the main blender process, isolating any data that might be overwriting by the player. Pthreads are used to render in parallel where multiple cores are available, operating on the same dataset in the same memory space, with data access protected by mutexes where appropriate.

_________________
BroadBlues On Blues BroadBlues On Amiga Walker Broad

SMP on AmigaOS is not possible without breaking compatibility or being slow.
But additional cores could be utilized by OpenCL for example.

@Tom01
Or offload work to cloud processing ; )
Coming next to the Amiga Family of computers : p

_________________
AROS Broadway - AEROS - Aminux - AmiCloud - indieGO! Appstore - AmiWallet - VAN lossless video codec - AMC Amiga media Center -KrypUnite - LibertyNet - MinX - amigaNX

@Hypex

Quote:

Hypex wrote: @Dandy Quote: Hmm - I use IBrowse 2.4 (on OS 3.9/WarpOS) since a decade or so and never noticed a bug. That's most likely beacuse it has no memory protection and clobbered memory would go unnoticed. AmigaOS3 and the unbderlying hardware didn't reveal hidden bugs programs had.

Hmmm - but what is a 'hidden bug'?
Can we call it a 'bug' at all, if it never causes any noticable problems, as long as you use it according to the original destination?

I mean - IB was designed for OS 3.x and on a classic OS 3.x system it runs flawlessly.

If problems arise when you run it on OS 4.x then bad luck - but it was never designed for OS 4.x and so I'm not surprised that problems can arise there.

From the OS 4 perspective it may look and feel buggy - but from the OS 3 perspective it is certainly stable.

So I think it is a bit unfair to give it the general label 'buggy'...

If you run i.e. an 50 year old, unmodified engine with lead-free petrol it might work flawlessly (in this case you are lucky), but it also might harm or even destroy the engine, as it wasn't designed and built to run with lead-free petrol. The same goes for old, unmodified Diesel engines, if you suddenly get the idea to run them with bio diesel. If you are lucky, they just run flawlessly. But they also might start leaking, as the gaskets become leaky by the use of bio Diesel.

Quote:

Hypex wrote: On OS4 is is unaccetpable as it has memory protection. At least to an extent. So any bugs in IBrowse will show up. The other thing is that it is running on an emulator. The emulator may have a bug. But either way that doesn't mean IBrowse is innocent becaue it doesn't crash on an OS3 Amiga. ...

I never coded for AmigaOS - neither for OS 3.x (or earlier), nor for OS 4.x. Furthermore I haven't had a chance to review the IB code.Therefor I cannot judge
if the code is buggy. But in favour of IB and it's devs I'd assume that it is stable as long as you don't change the conditions to run it. And this assumption is supported by my own experience - it runs flawlessly as long as you stick to the requirements.

If you don't stick to the requirements, you should not be surprised if strange things happen or it doesn't run at all.

Last edited by Dandy on 15-Aug-2014 at 07:03 AM.

_________________
Ciao

Dandy
__________________________________________
If someone enjoys marching to military music, then I already despise him.
He got his brain accidently - the bone marrow in his back would have been sufficient for him!
(Albert Einstein)

I wonder if blender will be one of the first apps to demonstrate AOS4.2 using multiple cores. @ amiwest 2014 perhaps.

"it just doesn't crash anymore with the native PPC port, but presumably the 68k 2.5 will be almost as crash happy as 2.4."
btw. what's the status of PPC native IB? Anywhere? How to get? Any plans?

Last edited by KimmoK on 15-Aug-2014 at 07:58 AM.

_________________
- KimmoK
// For freedom, for honor, for AMIGA
//
// Thing that I should find more time for: CC64 - 64bit Community Computer?

@itix

Quote:

There is no more memory protection in OS4 than there is in OS3

That doesn't explain why accessing location zero or going over a boundry crashes.

Quote:

In fact it is easier to spot illegal read/write access in OS3 because typically they have very little ram.

How does help? Most likely that access will be ignored by the hardware.

Last edited by Hypex on 21-Aug-2014 at 04:31 PM.

@Dandy

Quote:

Hmmm - but what is a 'hidden bug'?

A bug in code that is unknown. It may go unnoticed or it may show up at a rare oppertunity.

Quote:

Can we call it a 'bug' at all, if it never causes any noticable problems, as long as you use it according to the original destination?

Yes you can. I'll give two examples. The first is some code I was writing once. I misread the DOS API and ended up passing NULL pointers as strings. Without knowing I expected the OS to test the string to be valid before using it. Two things were wrong with that. One is that the API said it must be a pointer. Second is that the routine will actually read the pointer regardless. I tested my code on OS4.0 once and all seemed fine. But then after an update my code started crashing. I considered my code to be clean and by the book or at least close to. I was shocked how many violations it had after I showed a developer my source. It was like going over a document five times correcting spelling mistakes only to find out later you misssed ten.

My second example is more sensitive. I was porting an interrupt music routine to AHI. During the course of that I emulated the interrupt in user mode. I found out that the orginal code actually used as lot of uninitialised pointers, in an interrupt! Now that I found shocking.

So, yeah, even if you see no problems, doesn't mean it's all innocent under the hood. Memory could be trashed and you wouldn't know a thing abbut it.

Quote:

If problems arise when you run it on OS 4.x then bad luck - but it was never designed for OS 4.x and so I'm not surprised that problems can arise there.

No but if it is written correctly it shouldn't crash. Lots of 68K apps are fine on OS4. Wordworth is usually fine but can crash on OS4. That doesn't mean it has no bugs if it doesn't crash on OS3.

Quote:

So I think it is a bit unfair to give it the general label 'buggy'...

As I recall I said it could be IBrowse that is buggy or the emulator. But that is not all. It relies on MUI so that could also be buggy. Infact I find MUI programs tend to crash most of all because of some bug in a MUI class!

Quote:

Therefor I cannot judge

Code can be juddged to be clean or messy but I doubt an obvious bug will stand out after a browse of the code.

Quote:

If you don't stick to the requirements, you should not be surprised if strange things happen or it doesn't run at all.

If those requirements are met or surpassed and the code is OS friendly it should work flawlessly.

Okay another example. There is a DOS function to connvert a lock into a file handle. Unless you look at the fine print AKA the autodoc for the function you could miss that the lock will be freed on success. It's easy to assume you still need to free the lock. I would have. On OS3 this will breeze right by, at least on FFS. On OS4 this will freeze the system.

Last edited by Hypex on 21-Aug-2014 at 04:36 PM.

@Hypex

Quote:

Quote: There is no more memory protection in OS4 than there is in OS3 That doesn't explain why accessing location zero or going a boundry crashes.

Uhm, and you never crash OS4?

Quote:

Quote: In fact it is easier to spot illegal read/write access in OS3 because typically they have very little ram. How does help? Most likely that access will be ignored by the hardware.

If you have CyberGuard or similar software it can be reported to serial output before system crash.

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

@Dandy

Quote:

hanks for this information. NutsAboutAmiga might be interested in this.

Well its possible that JIT compiler breaks some thing that should be atomic, but it should not, propper synchronistic should be provided by native Exec functions for Mutex/Semaphore . I say the design is broken if they depend or hope some 68k assembler opcode is atomic. it might also be stack overflow issue. And it might be some thing totally different. It does not make mutch difference to me as a user. Can't make changes to code, and can't fix it, so the best thing to do is to avoid it.

In any case IBrowse might be faster, but its not noticable on AmigaONE-X1000, on slower Sam440 it might make sense to try to run it, but even then there NetSurf and its good contender to Ibrowse. there is also Aweb.

IBrowse is old and dose not support HTML5 nor CSS, renders most web pages wrong, and too anying for my taste.

Quote:

Just out of interest: I understood you were using the 68k version with OS4, right? What about the "limited OEM version of IBrowse that is included with AmigaOS 4"? How is this version regarding stability?

The same.

Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:02 AM.
Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:02 AM.
Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:00 AM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS

@Hypex

Quote:

Quote: There is no more memory protection in OS4 than there is in OS3 That doesn't explain why accessing location zero or going a boundry crashes.

If a string point to address 0x00000000, then it should be safe, shure you get DSI error because its defined as illegal memory access, but first byte on address 0x00000000, is a 0 (end of string), so your program continues to work, all the programs sees is empty string at address 0x00000000.

As long as the application does not try write “hello world sucker” at address 0x00000000, the OS and program will not crash.

Shure there might exceptions like if you did it in forbid state or in intuition lock, you might end up with a system freeze.

In any case if the program is not working correctly dont use it.

Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:18 AM.
Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:16 AM.
Last edited by NutsAboutAmiga on 22-Aug-2014 at 03:16 AM.

_________________
http://lifeofliveforit.blogspot.no/
Facebook::LiveForIt Software for AmigaOS