-=> On 11-01-95  22:54 you wrote to me <=-

Hi Paul,

PE> Thank you!  The bulk of stuff that I write will work on any ISO
PE> conforming compiler.  That includes my mailprocessor, Tobruk.

MS> Does this mean I can run a point, using code that compiles on my
MS> Amiga? That's great! (But I'll try to get Devil working first.)

PE> Yes, using the Devil Dialer with some slight mods, you can be
PE> up and running as a point to send/receive mail.  Using Tobruk
PE> as your mailprocessor, you can process the mail (but actually
PE> in the portable config it is only of use as a hub, since it
PE> takes PKTs and produces PKTs, it is implementation-specific
PE> what messagebase it writes to, and in the Amiga config it won't
PE> write to any messagebase).  What messagebase would you have 
PE> used anyway?

I don't know, as I said before, I don't know much about FIDO or
comms.

PE> I've got another program, PQWK245 which converts from PKT to
PE> QWK and vice-versa, but I didn't write it, just modified it,
PE> and it is not written properly portable (although it may work
PE> anyway, as I did manage to make it work on 32-bit OS/2 as
PE> well as 16-bit DOS + OS/2).

PE> That's interesting.  When I actually went to compile MD5, I found
PE> a couple of problems, which I fixed, and are included in the latest
PE> OZPD (can't remember when I made the change), but anyway, the 
PE> latest version of OZPD is OZPD9510.ZIP, available for FREQ as
PE> "OZPD" from 3:711/934.  I would be interested in any changes you
PE> needed to make for the Amiga.

MS> I didn't actually change anything, but I did need to create a
MS> config.h file from scratch.  I also created a test program, which

PE> Oh yeah, that's the crap from CVS.  I've gotten rid of that stuff.
PE> Try taking my modified version.

Well, the version I already have works, or seems to.  I'll have a
look at the modified version when I have the Devil Dialer working so
that I can Freq it.  :)

MS> prints the name & digest of all files named in the command line.
MS> This is useful to search for duplicate files.

PE> I'd also be interested if you actually understood the algorithm,

MS> Yeah, sure! :) I think I understand the general principles, though.

PE> and could take out any dependency on longs being exactly
PE> 32-bits.  A few & 0xFFFFFFFFUL should fix that problem, but
PE> someone actually needs to take the time to read the code to know
PE> where to put them.

MS> The code isn't that long.  :) I think that it might only require
MS> masking in byteReverse(), but I'll have a closer look soon.  OTOH,
MS> it might be better to get someone with a 64 bit environment to work
MS> on this.  :)

PE> It's actually got nothing to do with a 64-bit environment.

Oh. What would go wrong if longs weren't exactly 32 bits, then?
Maybe I'm not understanding the problem here...

PE> The fingerprint is actually 128-bits (I used to be under the
PE> misapprehension that it was 64 bits), but it is implemented in
PE> 32-bit chunks only.  Which of course the Amiga is.

PE> P.S. With a bit of detective work (the date that your doco said
PE> you modified your script, the dates of calls to my BBS, plus your
PE> unusual (2400) connect speed), I was able to take a fairly
PE> accurate guess at which connects were you.  Your script originally
PE> had a problem of sending x'd1' instead of x'f1'.  You obviously
PE> fixed this up pretty quick though.  However, because of your failure
PE> to send x'f1', my system logged a lot of extra information, which
PE> told me:

PE> 1. You are sending "DEVIL DIALER" in all-capitals instead of mixed
PE> case.

I am?  A very early version might have, but the latest attempts
shouldn't have.

PE> 2. You are not following the string with a whole lot of spaces.

PE> 3. You are not terminating the string with a NUL character.

PE> Having looked at your script, it looks like REXX doesn't preserve
PE> the spaces, and it doesn't bother to send NUL characters.  Wanna
PE> see if you can do something about that?

Rexx itself can cope with the spaces, etc, but this is a common
problem with poorly implemented ARexx command hosts - I think I know
how to get around this problem.  I think I'd better set up a
null-modem link to test this thing properly.

PE> If you can tell me an exact date + time (Sydney daylight saving
PE> time) that you want to call, I can set up Binkley to log every
PE> single character received and sent, so if my suggestions above
PE> don't work, try that out.

I'll have a close look at it soon. I'll wait till I get feedback
from the other night's efforts before arranging another experiment.

Michael Stapleton of Graphic Bits.

 * AmyBW v2.10 *
... This tagline is encrypted
--- Blue Wave/RA v2.10 AmyBW