[Coco] My Windows Readme

[Gene Heskett]

On Sunday 12 April 2015 07:33:26 Bill Pierce via Coco wrote:
> Stephen, with all your knowledge, wisdom, and superior intelligence...
> 'cmdgen' is NOT some renamed shellscript or Microware's 'build'
> command renamed. It is a machine language program written by S. B.
> Goldberg and is one of the many 'Goldberg Utilities' uploaded to
> Delphi, which is where I got it from. Steve Goldberg wrote MANY fine
> utilities and modules for OS9, some of which were published in Rainbow
> and other publications as well as sold commercially.

> As for the 'Sleuth' X-disassembler... The one on Gene's site works.
> Whether it has the 6502 code or not, I can't say,

It should be there Bill, but I never threw the switch to build it to take 
6502 code apart for the relatively lame excuse that I didn't have one of 
those half of one leg critters.  IMO, the 6502 only has half a leg 
compared to the 6x09's, the most limited architecture of anything I ever 
looked at the tried to claim it was a cpu. 

The sources for 6502's are there AFAIK, in the downloads I have showing.
If not, I made a mistake when I repacked the working version, but folks 
have had all the time since I setup that web page to let me know it was 
missing.

AFAIK, the only Senator Foghornish squawks are from SHF, always presented 
without an example that someone else could stand a chance to duplicate.

So, until such time as I SEE on my screen, by following the exact 
procedure that got him to conclude that its broken, I will defend it. I 
suspect its 99.9999% PEBKAC.

SHF went off on a shellplus siding years ago, got stuck on its deadend 
and has never found the switch to get back on the mainline.

> but the original 
> version is in the SWTPC files I linked to earlier but it is not 'Level
> 2' compatible (missing the L2 system calls)

Missing more than L2, its defines are pre-level 1, version two, which 
that $29.95 upgrade very early on changed them all by switching from 5 
byte strings for the names to 8 byte strings for the names.

> which is why Gene was 
> working on his version (with the original author's permission).

Which does use both the L1 and L2 defines by dissing code written and 
assembled with the original L1 5 byte call define names by disassembling 
that code using the exact equivalent of them in the later 8 byte format. 
They are 100% equivalent.

If there some L2 gfx calls missing I'm sorry, but since the whole thing 
is limited to 256 choices by the structure of the software IRQ in the 
6x09, the gaps in the table s/b pretty obvious, and they can be plugged 
right in at the proper "enum" place in the table, replacing the nulkl 
bytes there now, and then reassembled to bring them to life.

However, I haven't tried to build it with lwasm here on this amd64 box.0

That might be an interesting exercise for a younger mind.

> There 
> is still a little work needed to be done on how it sets up the header
> variables, but still better than most of the OS9 disassemblers I've
> tried and I've tried about every one in all the archives.

The one thing it doesn't do, and I suppose some would call that a header  
problem, is that you must write a file to direct how Sleuth interprets 
the data it find, address range by address range.  In other words, if 
there are a few stanzas of text as fcb/fcc or fcs strings in the code, 
it is up to YOU to tell Sleuth in this support file, what this data is.

Sleuth comes with extensive documentation.  And they are printable. READ 
them and follow the directions and it will work perfectly.

If you can't read and follow them, then you get to keep all the pieces.

All the other dissassemblers we have try to guess what the data its 
looking at is, byte by byte, and they all blow it.  They will usually 
spit out code that will reassemble to that same byte, but the resultant 
source file is indeed going to have some funkity looking nonsense in it. 
But they work MOST of the time and are a bunch easier/faster to use. 
Sleuth OTOH, gets it right if you are an experienced enough coder to 
write that data file for it.  Something that must be done for every file 
you sic it after. Thats why its slow, but you get perfect source code 
with that bit of parallel effort.

> Bill Pierce
> "Today is a good day... I woke up" - Ritchie Havens

[much snippage]

Cheers Bill, to you and the rest of our readers here, Gene Heskett
-- 
"There are four boxes to be used in defense of liberty:
 soap, ballot, jury, and ammo. Please use in that order."
-Ed Howdershelt (Author)
Genes Web page <http://geneslinuxbox.net:6309/gene>