> port MorphOS to MiniMig.

MorphOS for m68k? What for?

> I never got a legacy Amiga program to function. What am I doing wrong?

Which ones have you tried?

> ISA Shift request: Just port it to Efika MX Smartbook or equivalent device. :-)

There are ARM devices with multiple times the performance of the Efika MX, therefore I'd be against restricting it to equivalent devices :-)

@connor

It's strange that OWB crashes while having enough memory left. It rarely happens to me. Maybe you have a faulty RAM?

@all

Well, we have diskussed this before and it's safe to assume that the MorphOS Team is already thinking about what the best option for a ISA shift is. Remember, it's not just about what platform is the "best", but also which a small team like theirs can realistically do.

That said, I'm reserving money for the day I need to buy new hardware and a new licence. My bet is that regardless what hardware the Team choose to support, the end result will be AWESOME!

Quote:

Are you sure it's saying it's because of the memory needs but not for other reason? What does it say in the requester? Because I for example get similar requesters with "assertion failed" text after trashing the memory with SimpleMail/68k :) With a quick look it looks like out of memory requester, but it isn't :) Luckily we now have better MorphOS port of SM and I haven't seen those OWB requesters anymore...

Quote:

And I presume you already know my "imaginary polls' result"?

@ Yasu

No it's no faulty memory. It is working in all other situations when memory is almost full. I can fill it with all the other apps that I use. Only OWB makes these problems.


@ jPV

It also happens when I do not use 68k apps at all. It is definitely OWB. What it says? This one:
Failed to allocate 5411766 bytes.
Unfortunately, WebKit doesn't check allocations and crashes in such situation.
You can either:
- Crash
- Retry
- Quit

I even got this message only because I hit Ctrl+T to open a new tab!

Weird... I haven't seen that kind of behaviour on my machines. Only when memory really is used.

Try "memtest extensive" in shell just in case?

has already been done. the memory is not faulty, it is all working

@connor

Strange indeed. My G5 does not behave like that.

Quote:

As mentioned before, it could very well be that your largest block isn't enough, even for a simple new tab, because a blank page still involves memory hungry steps, in particular :
- allocating the cairo surface (but since i manage it myself, it's properly handled and it wouldn't crash if the allocation failed).
- instanciate webkit js heap stuff, which can take several megabytes.

THanks. I will watch the largest free block with this tool and report what it says. How much memory does it need for an empty tab? It is really a shame that for a simple empty page it can eat up megabytes of RAM just for showing nothing. This Cairo stuff is not good in my opinion. It seems to be so wasteful with resources. So all in all it means: there is no way out until we have a MOS with virtual memory and all that stuff.

Quote:

The cairo memory usage for a blank page is mostly defined by the webview area size, i.e widthxheightx4 bytes.
Anyway, using cairo is not the problem here. The real issue is the way JavaScriptCore reserves memory, as mentioned in earlier posts, which is equivalent to huge memory leaks to us, and there's not much i can do about it. I tried several approaches, without success (like clearing/freeing everything/reinstanciating a VM instance and so on). The only solution in our context (shared memory space) would be to have each view in its own process, since the JS virtual machine could be freed when the view is disposed. Unfortunately it's really quite some work.
In practice it means porting WebKit 2 and reworking many parts in Odyssey's internals.

I think going to WebKit2 would be good because then it should be easier to fulfill the requested memory sizes. So it should run better on our system. So we had a better control over the memory usage. But I can guess that it will be a lot of work and very boring for you because there is "not so much more" usage of it.