home .. forth .. colorforth mail list archive ..

Re: [colorforth] part two of Ray's post

Hello Colorforth!
Hi Nick, Ray here...

On 9/17/07, Nick Maroudas <alice@xxxxxxxxxxxxxxxxxxxxxxx> wrote:
> Quoting "Ray St. Marie" <ray.stmarie@xxxxxxxxx>:
> Fri, 14 Sep 2007 13:33:02 -0600
> >
> Nick here:
>
> Dear Ray,
>
> Thanks for the interest; a bootable floppy is on its
> way
> to your address.  It also includes a block of code that
> might enable you to play an audio CD thru your
> soundcard while coding ColorForth.
>

Much thanks, can't wait to render the bytecode into html, of
which I will post at my new site (will be new after next week.
colorforthRay.info.

I lost colorforth.info to a bot. :-( bonehead mistake, my talent. :-)

>
> http://www.strangegizmo.com/forth/ColorForth/msg00692.html
>
> http://www.strangegizmo.com/forth/ColorForth/msg01275.html
>
> http://www.strangegizmo.com/forth/ColorForth/msg01287.html
>
> http://www.strangegizmo.com/forth/ColorForth/msg01485.html
>
>

Thanks for these links! yummy to eat with breakfast.

> 1. In reply to your suggestion of burning the floppy
> image into a cd, this is not reliable on my hardware.
> For instance, I find that the floppy version of
> Howerd's
> CFDOS4 boots reliably, but sofar have not succeeded
> with its CD image by John Comeau - pollution in the
> electric mains here is dire.
> >

I was suggesting that the floppy image remain the floppy
image so that the disk would pass thru magnetic fields
incorruptible, not so that I could run it from the CD.
Lets hope the image travels well. If not, I'll have
cable on Wednesday...

> 2. As regards trying the 12bit parallel DAC from Analog
> Devices, they alse make a 16bit board ($150) which I
> hope to drive at 1 MHz with a PCI proto board plus fast
> 32bit cable from Quancom.de (300 Euro).  I am going to
> all this expense and trouble because of a notion that
> really accurate tuning (not semitones but 1/8th tones)
> needs  faster conversion rates (about 10 times faster)
> than current commercial codecs can supply.  Also,
> audiophiles seem to agree that parallel conversion is
> less jittery than the present wave of serial
> convertors.
> >
>

I need to learn so much... That's totally cool! When I can
afford a bit, I'll be pumping you constantly for how to's on
getting up and running. Might be a nice feature on the
new .info site. :-)

> > 3. As to the AC-97 block included in your Multi tasker
> image with Tim Neitz, I have a printout of the same code
> (plus a lot of web code) that I always thought came from
> Chuck via Tim. If so, I also remember Chuck writing in
> an early CF post, that his card needed to be
> initialized by an external OS before switching to CF.
> Has anyone on  CF found out what that first step is -
> for any card?

Not I and this is new info for me as well. Thank you,
more to explore.

> That is the point at which I got fedup and went the
> parallel route.  But AC'97s are cheap and good for most
> music purposes, and I guess from the blocks Tim so
> thoughtfully provided (on interrupts also) that we have
> almost enough to code a complete CF audio driver:)
> >
> >

That's very good news as well. Must look into some of the
Linux drivers and make a list of what's available.

> >4.  You say you'ld like to go over 05 again.  Well, the
> floppy is Josh's 16bit CF05 with some helpful macros
> that I snipped from Howerd's CF4DOS. I also amended
> block 30 to correct the color balance, due to a little
>
> difference between 32bit and 16bit color coding (see
> Karig's commentary).
> >

Okay, this is where I have a little to share with you...

In the '05 I've found two of the problems I was a tad
concerned about, being that Josh has never run
the system, and wasn't aware of the operation of it.

1) WARM -- this appears to be coded slightly
incorrectly in '05 compared to every other version
I am looking into.
I wasn't sure that this was a change or a bug, but
I'm leaning bug at this point. It could be a change,
as the '05 system seems to start with MAIN (me)
task rather then GOD, but I'm not convinced

The problem is typing WARM crashes the system.
I'm thinkng the solution is to be sure to allow WARM
to drop thru to the ati0 call.

Example --

'05 problem code... not the entire, just relevant...

warm:
	DUP_
	jmp short start1.0
start1:
	call ati0
.0:	call noshow
	call noserve
	mov dword [forths], kforths
	mov dword [macros], kmacros
	mov dword [trash], buffer
	; copy initial forth dictionary up to its real address
( omission -- to end of code labeled warm)

Original Chuck Code, jul 22 2001

warm: DUP_
start1: call ATI0
; omission... two lines commented out
    call show0
    mov  forths, (forth1-forth0)/4
    mov  macros, (macro1-macro0)/4
    mov  EAX, 18
    call LOAD
    jmp  ACCEPT

It's entirely possible that the need to jump to the Noshow(show0)
has something to do with the new starting place, ME in '05 and
GOD in everything else... But I doubt it. I'm certain this is why
'05 won't WARM boot properly. I havn't tested this yet. I will.

ALSO>>>

The color problem...

> block 30 to correct the color balance, due to a little
>
> difference between 32bit and 16bit color coding (see
> Karig's commentary).

Original Chuck contains a routine called label: rgb that converts
hex-color numbers into the required form, omitted from the '05.

rgb: ror  EAX, 8
    shr  AX, 2
    ror  EAX, 6
    shr  AL, 3
    rol  EAX, 6+5
    and  EAX, 0f7deh
    ret

This routine makes porting the block 30 easier between versions,
I'm assuming.

I'd like to test these ideas as soon as I can.
I am fortunate to have Chris R Walton, ( arke in c4th chats)
coming to visit sometime next week, as he's home from
school in Germany and visiting parents that live the
next town over from where we have moved into.

We have plans to 'hook-up' and I will show all of this
to him for his opinion.

I'm hoping Josh Grams is still receiving this mailing list, and
will pipe in as necessary. He's busy on his farm and we don't
seem to be online together as often any more, being that
I'm dial-up until next Wednesday.



In other news...

Ya I did lose the name colorforth.info to a bot. Looking at
WhoIs data for my renewal date, I assumed that I would be
good to go until July of 08, as that was the reported
renewal date in the Whois. I was wrong, and by the time
I figured it out, was over 70 days over due for payment.
Hostway claims there was nothing they could do to insure
that I got the name back, as I found out about this prior to
the names release back into the public... I lost it, my fault.

Currently the old page loads as ColorforthRAY.info.
I'm in the throws of updating the site. When cable arrives,
I'll start uploading the new stuff.

I still plan to make it an excellent resource, so long as I
don't allow other forthers, many of whom I respect,
throw me off my colored horse again.



Nick I can't wait to digest your code. We don't get a
lot of fresh stuff to read. :-) Also, I can't wait to
get it running and making music with me.
That's a lifelong goal and I thank you for the
work you've done.

I want to find a way to jam online in real time if
somehow that is possible, and to allow for a
distributed band and audience.
Might be the next Broadcast dot com. lol :-)

>

> >Caritas,
>
>            Nick

My great pleasure is conversing
about these things with you.

With Respect,

Ray
-- 
Raymond St. Marie ii,

---------------------------------------------------------------------
To unsubscribe, e-mail: colorforth-unsubscribe@xxxxxxxxxxxxxxxxxx
For additional commands, e-mail: colorforth-help@xxxxxxxxxxxxxxxxxx
Main web page - http://www.colorforth.com