Re: [MiNT] 32-bit FreeMiNT kernel

[Paul Wratt <paul.wratt@gmail.com>]

On Wed, May 19, 2010 at 2:45 PM, Jo Even Skarstein <joska@online.no> wrote:
> On 05/18/2010 10:47 PM, Miro Kropacek wrote:
>
>> just an idea to discuss -- while hacking CT60 TOS I realized we can do
>> now things which seemed totally impossible in the past like native
>> FAT16 / FAT32  support, hard disk driver inside and voila -- 32-bit
>> TOS API which would make possible, among others, also to make FreeMiNT
>> kernel 32-bit (i.e. able to compile without -mshort). Of course, we
>> have to keep compatibility for old apps, so new parameter on stack
>> would be introduced for new apps (in mintlib and freemint) so that
>> 32-bit API is used.
>>
>> I just wanted to know your opinion on this, is it any good, except
>> good feeling we're 32-bit? :)
>
> What would the benefits of a 32 bit API be? As others have pointed out,
> we've always had 32 bit pointers, so I think the speed improvements
> would be marginal.
>
> I'm not so sure if it's worth the effort. Wouldn't mind embedded FAT32
> and harddisk driver though :)
>
> Jo Even
>
The embedded drivers would be nice.

A question:
what is the probability of large TOS roms running on current hardware

If it is, the added ext2 and usb support would be a must (presuming
current TOS roms are already cramped)

a debug TOS may be useful too

just to go left field, what are the possibility of:
A) TOS containing EEPROM burning (for updating)
B) TOS being split into ROM/EEPROM & static ram (for partial dynamic
updates or changes)

here I am implying BIOS type use and behaviour. static ram being
changeble in the way EEPROM is not (sorry can't remember the proper
description or name)


Paul