Re: [MiNT] XaAES sources for FreeMiNT 1.16.3

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

On Mon, Nov 30, 2009 at 12:45 PM, josephus <dogbird@earthlink.net> wrote:
> Peter Slegg wrote:
>>
>> On Mon, 30 Nov 2009 06:08:54 , Paul Wratt<paul.wratt@gmail.com>  wrote:
>>
>>>
>>> maybe but that doesn't explain why there isn't a problem when running
>>> 1.16.3
>>>
>>> just out of interest, has anyone trying compiling current XaAES
>>> against 1.16.3 kernel, or 1.16.3 XaAES against current kernel
>>>
>>>
>>> Paul
>>>
>>>
>>
>> On the Milan I have 1.16.3 from May 2007 with XaAES 0.998 from
>> 6th May 2007. I think these were the last releases. It's been
>> that long.
>>
>> Rather than compiling everything, why not use these ?
>>
>> Peter
>>
>>
>> I was remembering  STONX  it uses  a swap algothim to map
>>
>
> Motorola to  Intel  and vice versal   would this help the convesation a
> littile?
>
> josephus
>

maybe, but if you look at the code, and read Lonny's description of
what he did, its very basic, and if you place enough "test" in then it
can be "fool proof" by default.

The wierd part part is, that with the same piece of hardware, 2
different results are being returned. If there is a "possible" problem
with current "hardware", then it may also interfere with future
developments.

I may have STONX source here, so I might have a look at that algorithm
you mentions, but honestly, I dont think its the detection routine
that is the problem (its has not changed between differing kernels),
its the data that is coming back to the detection routine which has
changed.

That does not mean that the detection routine doesn't need updating,
as a "better" (more thourough) pixel test would also "fix" the
problem. Howevert does show that there is currently an "unknown flaw"
(feature) with the current CVS FreeMiNT/XaAES combo, and that someone
should at least take the time to find out why or what it is..

Paul