Sub: Minutes of your RT Conference.

>From Steno: Jennifer Marlowe

Date: 05/07/89   Time: 21:21EDT
Attendees:
[[Dennis] D.RUFFER]
[[jax] FIGCHAPTERS]
[M.MELLEN1]
[[Jack of BCFB] J.BROWN11]
[[Wil] W.BADEN1]
[D.GOLDING]
[[Paul] P.THUTT]

Minutes:
<FIGCHAPTERS> is here.

<D.RUFFER> hi jax  {call the cops :-)}

<[jax] FIGCHAPTERS> Yo! You there? Musta gone to drink a glass o' water  hi

<D.RUFFER> no just setting up the windows again

<[jax] FIGCHAPTERS> So ... let's discuss whether the ANSI standard should
          be shot or hanged!

<D.RUFFER> why do you want to hang ANSI

<[jax] FIGCHAPTERS> Well, because they are mind-f***ing us, to use an old
          Hippie/Yippie phrase

<D.RUFFER> how so?

<[jax] FIGCHAPTERS> Well, they have everyone convinced (falsely) that
          publishing the BASIS entails legal dangers.  Another example ...
          of how out of touch they are with reality ...  anyone who has
          seen modern telecomm disclaimers ...  like: ...  "HEY YOU! YA,
          YOU! WE TAKE NO RESPONSIBILITY WHATEVER FOR THIS! FORGET IT !"
          would realize what a load of horsefeathers they have sold us.  .

<D.RUFFER> ok  well, in a way you are right...  ANSI themselves are a
          little behind the times...  but they also have been burned many
          times before...  they deal with many standards committees...  and
          many groups of users like us...  they have just learned to be
          very cautious  .

<M.MELLEN1> is here.

<D.RUFFER> hi m.m

<M.MELLEN1> hi

<[Dennis] D.RUFFER> what you do m.m (what is you firt name anyway :-)

<M.MELLEN1> first name is mike. I write s/w for Harris Semiconductor -
          RTX2000

<[jax] FIGCHAPTERS> Hey Mike ...

<M.MELLEN1> Yah?

<[Dennis] D.RUFFER> oh yea, I thought I recognized the name

<[jax] FIGCHAPTERS> I hear rumours that Software at Harris is an "orphan"
          department.

<M.MELLEN1> Well... we're in a strange position. we !have! to produce s/w
          because the  cucustomers "expect" it, but we don't try to make
          our products overly competitive with  anyone else's. Some
          incentive to do good, huh?

<[jax] FIGCHAPTERS> well, that's life. No incentive, huh? How about job
          security?  hah.  ..:-)  .

<[Dennis] D.RUFFER> sounds like the way harris is with all their products

<[jax] FIGCHAPTERS> Well, Dennis, I have met their chip theorists and they
          are damned sharp!  Harris has a young face.

<[Dennis] D.RUFFER> nice military products, but not very competitive on the
          commercial market

<[jax] FIGCHAPTERS> The prices the gov't pays, who needs commercial?

<M.MELLEN1> we're trying to change ouour Our purchase of GE,RCA, etc should
          change our commercial image

<[jax] FIGCHAPTERS> Tandy sells the TRS-80 Model 4 for $1100 these days!
          Because NASA buys them. A 128k Z80. Can you imagine?

<[Dennis] D.RUFFER> you are buying GE????

<[jax] FIGCHAPTERS> I'm all for Harris ...  I've been to their
          presentations ...  and they say "Forth!" loud and clear without
          blushing.  .

<[Dennis] D.RUFFER> I knew GE bought RCA, but Harris chews up GE....  that
          is a mouthful.

<M.MELLEN1> GE, Intersil, and RCA semi operations. You should see our new
          logo. It almost  doesn't fit on a business card

<[Dennis] D.RUFFER> ok, GE must have gotten rid of that business line

<M.MELLEN1> that was their orphan operation

<[jax] FIGCHAPTERS> Mike ... Harris called me about the RTX 32-bitter. They
          said they were going to come around and ask what we'd like to see
          in them.  Have they started that yet?

<[Dennis] D.RUFFER> I was just curious if GEnie got swallowed without me
          noticing

<[jax] FIGCHAPTERS> NO, GEISCO and the rest of GE are still firmly in GE
          hands.

<M.MELLEN1> the 32-bitter (RTX4000) is still in its infant stages. You will
          be hearing from us in the near future

<[Dennis] D.RUFFER> hey, you guys know anything about sorting algorithms?

<[jax] FIGCHAPTERS> Call me bubblehead!  .

<[Dennis] D.RUFFER> boo hiss, not the bubble

<[jax] FIGCHAPTERS> There's lots of sorts of sorts. What sort do you want?

<[Dennis] D.RUFFER> I've been playing with a KWIK sort algorithm
          recently...

<[jax] FIGCHAPTERS> Mike ... that's great

<[Dennis] D.RUFFER> in fact, it is the topic tonight...

<[Dennis] D.RUFFER> I finally got it working about a half hour ago...  you
          guys interested in seeing it?

<M.MELLEN1> sure

<[jax] FIGCHAPTERS> likewise

<[Dennis] D.RUFFER> ok

<[jax] FIGCHAPTERS> LISTING ok  .. :-)  .

<[Dennis] D.RUFFER> well, we are going to delve a little bit into polyForth
          to do this...  that is the language I'm using right now, and this
          is...  as I said, hot off the presses.

<[jax] FIGCHAPTERS> PAGE

<[Dennis] D.RUFFER> again, excuse the slight delays, I'm cutting from the
          source blocks...

<[jax] FIGCHAPTERS> GOLDEN

<[Dennis] D.RUFFER> ( Sort Utilities )
<[Dennis] D.RUFFER> VARIABLE COMP       VARIABLE TRADER     VARIABLE SORTER
<[Dennis] D.RUFFER> VARIABLE KEPT       VARIABLE KEEPER
<[Dennis] D.RUFFER>
<[Dennis] D.RUFFER> ~ : COMPARE ( n1 n2 - -1 | 0 | 1 )   COMP   @EXECUTE ;
<[Dennis] D.RUFFER>   : TRADE   ( n1 n2)                 TRADER @EXECUTE ;
<[Dennis] D.RUFFER>   : KEEP    ( n)                     KEEPER @EXECUTE ;
<[Dennis] D.RUFFER>   : SORT    ( n)                     SORTER @EXECUTE ;

<[Dennis] D.RUFFER> I see I'm going to have to shorten my name

<[jax] FIGCHAPTERS> is the ~ significant or is that a typo?  take it
          @EXECUTE is PERFORM  .

<D.RUFFER> these are the variables used by the system...  the ~ means to
          store the entire name characters...  to prevent collision with
          COMPILE...  which has only a 3 character name...

<M.MELLEN1> has left.

<D.RUFFER> did I bore Mike?   Do you want to continue Jack?  oh Jack, did I
          scare you away too?  is anyone out there?????

<[jax] FIGCHAPTERS> hi  eating, fingers greasy  ga

<D.RUFFER> is this too much?   Shall I go on?

<[jax] FIGCHAPTERS> ga, iwalk away & return  excuse, dinner & baby  ga

<D.RUFFER> ok, Jennie and I will talk while you eat :-)  BTW, @EXECUTE is
          F83's PERFORM...  but it adds a test for 0 and is a noop if the
          variable...  doesn't contain anything   (nice touch)

<[jax] FIGCHAPTERS> what

<D.RUFFER> the test for 0

<[jax] FIGCHAPTERS> if it contains ' @EXECUTE :-)

<D.RUFFER> no, that won't work....  it would be ' EXECUTE ....

<[jax] FIGCHAPTERS> aha  right  i had a funny Forth adventure  but goahead
          w/ SORT

<D.RUFFER> these would be used as ' COMP !  then you type COMPARE to
          execute it...  also, there is another "trick word" that I am
          about to use...

<D.RUFFER> : S@ ( # - b)   SCR @ BLOCK + C@ ;
<D.RUFFER> : S! ( b #)     SCR @ BLOCK + C! ;
<D.RUFFER> : TRD   TRADER ASSIGN  2DUP S@  SWAP S@  ROT S!  SWAP S! ;
<D.RUFFER> : KP    KEEPER ASSIGN  S@ KEPT ! ;
<D.RUFFER> : CMP     COMP ASSIGN  KEPT @  SWAP S@  - ;
<D.RUFFER>
<D.RUFFER> 1 +B LOAD ( Load sorting algorithm ) KWIK TRD KP CMP

<D.RUFFER> these are just some test words, but illistrate the word
          ASSIGN...  is is a little bit of a trick...  it stores the
          interpreter pointer into the variable, and then exits...  as you
          can see in the last line, you just execute the words...  and the
          execution vectors are set up...  the equivalent in something like
          F83...  would be to take out the variables and the ASSIGNs from
          the...  definitions, and do a ' on the words and store them into
          the variables...  this is just a little bit of a short cut...
          and allows me to redefine these words to do what I really want to
          do...  obviously if I write a sort, I want to use it to do real
          work...

<J.BROWN11> is here.

<[jax] FIGCHAPTERS> hijack!

<J.BROWN11> HI DENNIS

<D.RUFFER> well I jack...

<[jax] FIGCHAPTERS> what's your BITNET addr?

<D.RUFFER> you snuck in there on me...

<[jax] FIGCHAPTERS> Dennis ...

<D.RUFFER> I was just rambling on about sorts  yes?

<[jax] FIGCHAPTERS> Does FORTH Inc. still do standalone mini systems?

<[Jack of BCFB] J.BROWN11> figgy jax are you picking up the FIG conference
          yet on RCFB

<D.RUFFER> they will do anything they can find someone to pay for :-)

<[jax] FIGCHAPTERS> Yes, JB! Indeed!

<[jax] FIGCHAPTERS> Dennis, do they have a workstation Forth off shelf?

<[Jack of BCFB] J.BROWN11> did you get the little report that bc fig
          started up again

<[jax] FIGCHAPTERS> with PCB CAD?  Yes, JB! Indeed ...

<D.RUFFER> it depends on what you call a mini right now...  what kind of
          workstation you talking about?

<[jax] FIGCHAPTERS> I was QUITE impressed with oyour report ...  and am
          replying you on ForthNet.  What kind yu g, h

<[Jack of BCFB] J.BROWN11> do you think we might get other tellecommers to
          make reports

<[jax] FIGCHAPTERS> ps  oops. parity  what kind you got, dennis?

<W.BADEN1> is here.

<[jax] FIGCHAPTERS> JB , I garbled you, say gain?  Hi Wil!

<D.RUFFER> I'm not sure what you mean...

<[Jack of BCFB] J.BROWN11> sorry about butting in looks like there are two
          simultaneous threads

<D.RUFFER> hello wil!

<[jax] FIGCHAPTERS> Yes, no problem. Now eight threads.

<[Jack of BCFB] J.BROWN11> wow hi will,  jack b. is here too

<D.RUFFER> called interactive fiction jack :-)

<[jax] FIGCHAPTERS> Dennis, we need PCB CAD. Possibly new computer. Is
          there a shelf pF product?

<[Jack of BCFB] J.BROWN11> wil I have been posting you puzzles in the
          network Peer review conference

<D.RUFFER> not really, you are just talking about a terminal aren't you?

<[Jack of BCFB] J.BROWN11> i'm getting confused

<[jax] FIGCHAPTERS> If you buy a SUN with SUN OS ...

<[Wil] W.BADEN1> Jack - how do I get'em?

<[jax] FIGCHAPTERS> you get an op system and all sorts of software ... like
          PCB CAD

<[Jack of BCFB] J.BROWN11> I want to talk to wil baden,  let's go to anothe
          room wil baden

<[jax] FIGCHAPTERS> Does pF have ... standalone multiuser for like a 386 or
          68020 ...

<D.RUFFER> and you get UNIX, but good like finding a decent Forth

<[Wil] W.BADEN1> ok lead on.

<[jax] FIGCHAPTERS> JB, WIL ... use /CHA <room#> command

<[Jack of BCFB] J.BROWN11> will type: /roo 2

<[jax] FIGCHAPTERS> and PCB CAD?

<[Wil] W.BADEN1> has left.

<[Jack of BCFB] J.BROWN11> ops wil baden type /roo 2  has left.

<D.RUFFER> pF has many varieties of 386 and 68K Forths, but nothing on UNIX
          yet

<[jax] FIGCHAPTERS> Don't NEED Unix.  Want reason NOT to buy Unix.

<D.RUFFER> then why buy SUNs

<[jax] FIGCHAPTERS> Want multiuser multiprogrammed ForthStation with
          PCBCAD.  Just mentioned for comparison.  .

<D.RUFFER> why not MSDOS CAD system?

<[jax] FIGCHAPTERS> They ucksay.

<D.RUFFER> why mutiuser, when PCs are soo inexpensive?

<[jax] FIGCHAPTERS> There is a lot of Nike-netting in our joint.  It's time
          to put a stop to it.  A 386 runs System V.

<D.RUFFER> Nike-netting???

<[jax] FIGCHAPTERS> what's pF's answer?

<D.RUFFER> they don't have one right now...

<[jax] FIGCHAPTERS> ( Nike-net ... you run around in your Nike sneakers
          with disks in your hand)

<D.RUFFER> I ran into the same problem with the job I'm on...  they wanted
          MAP on a VME bus...  Motorola only offers it on UNIX  we could't
          offer them anything but putting in a UNIX system :-(

<[jax] FIGCHAPTERS> If Forth is to survive in desktop computing ...  We
          need some Forth-based minis.

<D.RUFFER> I agree, and I will mention that you are asking...  but don't
          hold your breath :-(

<[jax] FIGCHAPTERS> I'd love to do it myself.  I MAY do it myself, if I can
          get a cheap mini.  Amiga is nice for that, but not popular
          enough.

<D.RUFFER> I would prefer a WINDOWS version myself, but the I don't like...
          sharing my computer with anyone but myself

<[jax] FIGCHAPTERS> But maybe!  You are right ...  but there are reasons
          multiprogrammed systems exist.

<D.RUFFER> there are????

<[jax] FIGCHAPTERS> And Forth is pretty much a no-show in that field.  Yes,
          there are!

<D.RUFFER> what are they?

<[jax] FIGCHAPTERS> Source code control is a darn good reason for it.

<D.RUFFER> I can do that with a network!

<[jax] FIGCHAPTERS> Foof. Networks are as much work to maintain as
          multiuser.

<D.RUFFER> I had just such a system set up back at my last employer

<[jax] FIGCHAPTERS> BUT I approve of networks.  I would just be more
          comfortable with a 4-user mini.

<D.RUFFER> not if you buy a good one, but I agree, they take a network
          administrator

<[jax] FIGCHAPTERS> There is an IBM System 34 up for tax auction here in
          Golden next week.  It is outdated and a piece of you-know but I
          may bid. It is two-user.  .

<D.RUFFER> minis served a purpose back when computers where expensive, I
          just don't see the need any more  I used to work on a 34...

<[jax] FIGCHAPTERS> That's because all the minis are FROM the era ...  TELL
          ME ABOUT 34!  ga  .

<D.RUFFER> no more poser than a Z80 I thought  make that power

<[jax] FIGCHAPTERS> cp?  cpu?  .

<D.RUFFER> yes, they don't know much about mutitasking yet...  they started
          that design with the 32...  which was just a 8 bit computer (or
          so it seemed)...  then they added some memory and some disk...
          then designed a time slicer, and called it a mini  is Monitoring.
          I'm monitoring 2, but can't talk to them :-)

<[jax] FIGCHAPTERS> hmmm  so  waht is 34 cpu?

<D.RUFFER> have no idea jack, but it doesn't have much horse power...

<[jax] FIGCHAPTERS> oh well  .

<D.RUFFER> 2 users is about all you will be comfortable with  you would do
          better buying a 386 at 25Mhz

<[jax] FIGCHAPTERS> the thing about minis now ...  is that they are
          parallel machines  the WELL just sold the VAX ...  and bought a
          Sequent with 6 processors!  FLIES with 30 users on line!  .

<D.RUFFER> not Monitoring.  boy monitoring 2 was just too confusing

<[jax] FIGCHAPTERS> wha'ts happening ther?

<D.RUFFER> yes, now you are talking about a machine that is too expensive
          to dedicate to one user

<[jax] FIGCHAPTERS> is Monitoring.

<D.RUFFER> wil is having fun learning how to use GEnie...normal stuff :-)

<[jax] FIGCHAPTERS> not Monitoring.  let's go over ther!  has left.

<D.RUFFER> jack, that was the reason minis were "invented"  ah well, let's
          see if I can  is Monitoring.

<[jax] FIGCHAPTERS> is here.

<D.RUFFER> welcome back jack...  I'll keep monitoring what's going on over
          there

<[jax] FIGCHAPTERS> do you know what dan quayle listens to on his sony
          walkman?

<D.GOLDING> is here.

<D.RUFFER> what jack?

<[jax] FIGCHAPTERS> left, right, left, right ...  ... :-)  .

<D.RUFFER> :-)  hi don?

<D.GOLDING> HI ALL

<[jax] FIGCHAPTERS> hi don  .s

<[Wil] W.BADEN1> is here.

<[Jack of BCFB] J.BROWN11> is here.

<D.GOLDING> hi Will!

<[jax] FIGCHAPTERS> welcome back to civilization.

<D.RUFFER> welcome back jack and wil

<[Wil] W.BADEN1> hi don.

<D.RUFFER> sorry, I just was stuck in here

<[Jack of BCFB] J.BROWN11> hi what's the topic

<[Wil] W.BADEN1> what trubls with sorts?

<D.RUFFER> ah the topic....  are we ready to continue where I left off?

<[jax] FIGCHAPTERS> it's 7:30 MDT. I gotta go.

<D.RUFFER> I soved the problem about 8:00 tonight

<[Wil] W.BADEN1> No, i dont know where you were.

<D.RUFFER> ok, see ya jack

<[Wil] W.BADEN1> bye jax.

<[jax] FIGCHAPTERS> bye all, see you in January, Wil! :-)

<D.GOLDING> bye Jax

<[jax] FIGCHAPTERS> has left.

<D.RUFFER> I had just done the introductions  I was just about to show the
          real code...  I could still do that, and explain whatever
          questions come up...  about the undefined words...  anyone game?

<[Wil] W.BADEN1> ga

<D.RUFFER> not Monitoring.

<[Jack of BCFB] J.BROWN11> ga

<D.RUFFER> ok, here we go

<D.RUFFER>    ( KWIK Sort Algorithm )
<D.RUFFER> : DOWN ( # - #')  BEGIN  DUP COMPARE  0> WHILE  1+  REPEAT ;
<D.RUFFER> : UP   ( # - #')  BEGIN  DUP COMPARE  0< WHILE  1-  REPEAT ;
<D.RUFFER>
<D.RUFFER> : EXCHANGE ( #1 #2 - #1' #2')   BEGIN  SWAP DOWN  SWAP UP
<D.RUFFER>       2DUP > NOT IF  2DUP TRADE  SWAP 1+  SWAP 1-  THEN
<D.RUFFER>       2DUP > UNTIL ;
<D.RUFFER>
<D.RUFFER> : EOL ( # n1 n2 - ... #')   SWAP 2>R  BEGIN  I I'
<D.RUFFER>       2DUP + 2/ KEEP  EXCHANGE  R> ROT >R  SWAP 2DUP <
<D.RUFFER>       IF  ROT 1+  ELSE  2DROP  THEN
<D.RUFFER>       I I' < NOT UNTIL  2R> 2DROP ;
<D.RUFFER>
<D.RUFFER> : KWIK ( n)   SORTER ASSIGN  0 SWAP 1-  1
<D.RUFFER>    BEGIN  1-  ROT ROT EOL  ?DUP 0= UNTIL ;

<D.RUFFER> ok this probably needs a little expaining...

<[Wil] W.BADEN1> no.

<D.RUFFER> first the ASSIGN word, do you all know what it does?

<[Wil] W.BADEN1> What's the problem with it.

<D.RUFFER> nothing any more...

<D.GOLDING> has left.

<D.RUFFER> I finally found my flaw :-)  I was not saving the compare point
          with KEEP...  thus the point for the COMPARE in UP and DOWN was
          changing on me...

<[Wil] W.BADEN1> I presume ASSIGN is polyForth ASSIGN.

<D.RUFFER> yes it is wil

<D.GOLDING> is here.

<D.RUFFER> welcome back don  do you guys mean you understand this mess?

<[Jack of BCFB] J.BROWN11> Sorry, but the barbeque is calling

<D.RUFFER> ok see ya jack

<[Wil] W.BADEN1> BYE JACK.  bcnu.

<[Jack of BCFB] J.BROWN11> it is hot up here and gotta go for supper  will
          check back in an hour..

<D.GOLDING> hi again, I thought tonight was going to be VP Planner, so I
          went to the other rom...

<[Jack of BCFB] J.BROWN11> has left.

<D.RUFFER> no, VP Planner will be on this thrusday  oops thursday

<D.GOLDING> OK  typos are ok in reattime!

<[Wil] W.BADEN1> Let me look at the coding offline. the idea's look right.
          ...

<D.RUFFER> that sort will be a little clearer if you download the steno
          notes after I post them next week <grin>

<[Wil] W.BADEN1> Did u write the code from scratch? or adopt from
          somewhere?

<D.RUFFER> it is designed to be mutipurpose, and just something to play
          with  no, I adopted from a file we have here on line...  the
          files is called VISISORT.ARC  HERE IS THE AD FROM SCREEN 0

<D.RUFFER>        VISIBLE SORT ROUTINES
<D.RUFFER>
<D.RUFFER>             taken from
<D.RUFFER>
<D.RUFFER>      COMPUTE'S GAZETTE - MAY 1985
<D.RUFFER>
<D.RUFFER>         "Understanding Sorts"
<D.RUFFER>                 by
<D.RUFFER>            Arieh Shamish
<D.RUFFER>
<D.RUFFER> : =================================:
<D.RUFFER> : Converted to FORTH by Norm Arnold:
<D.RUFFER> : =================================:

<[Wil] W.BADEN1> It's a shame that it will crash your system once in a very
          rare while.

<D.RUFFER> he did a bubble sort and a shell sort also...  oh yea will?
          Under what cercomstances

<[Wil] W.BADEN1> You shud add a line to choose the smaller of two halves.

<D.RUFFER> I'll play along :-)  why so wil?

<[Wil] W.BADEN1> so that u can guarantee that the required stack size ...
          will not exceed log(base 2) of N.

<D.RUFFER> yes  it will fill up the stack pretty quick...  I havn't
          calculated how much yet.

<[Wil] W.BADEN1> Otherwise a very rare ...  combo of values will cause
          stacking ...  for every pivot.  I will look at it offline and
          give a full report. ga

<D.RUFFER> ok, have you used the KWIK algorithm before?

<[Wil] W.BADEN1> This is quicksort by c.a.r.hoare, published in 1962.

<D.RUFFER> thank you for the precise reference

<[Wil] W.BADEN1> wait a moment...  Collected Algorithms of A.C.M., #63 and
          64.  1961 not 1962.

<D.RUFFER> ok, it sound like I'd better put in a stack check them.  I was
          considering that anyway

<[Wil] W.BADEN1> this is the first step in finding ...  OSwhat is still the
          fastest known algorithm for sorting ...  on a single processor.

<D.RUFFER> does that reference contain some better algorithms?

<D.GOLDING> what is the fastest?

<[Wil] W.BADEN1> over a period of several years it was improved
          considerably.  OSthe basic algorithm is good but more
          sophistication in choosing ...  OSthe pivot and optimizing the
          actions.

<D.GOLDING> is their a limit to the number of items sorted?

<[Wil] W.BADEN1> No limit.

<D.RUFFER> this is limited by stack size

<D.GOLDING> that is my question.  there is a limit?

<[Wil] W.BADEN1> As long as the stack is at least log(base2) of N  (times
          two).

<D.RUFFER> each "pivot" as wil calles them causes 2 items to be put on the
          stack

<[Wil] W.BADEN1> Thus to sort 1,000,000 items requires a stack depth of 40.

<D.GOLDING> I have 100,000 items, how large a stack do I need?  ok  how can
          you be having stack size problems with 40?

<D.RUFFER> actually, I also have to have room for about 5 additional items
          on a single iteration  I didn't have any stack problems yet...
          but haven't done anything of significance with it yet.

<[Wil] W.BADEN1> stack problems are very very very very rare but can happen
          ...  and when it does will be a complete mystery.

<D.GOLDING> I thought you're system would crash sometimes....

<D.RUFFER> no, I had a basic algorithm problem until about 8:00 tonight
          <grin>  that was when KEEP crept into the system :-)  it makes
          sure I can keep track of the pivot point

<[Wil] W.BADEN1> You need 34 stack depth plus a few for scratch for
          100,000.

<D.RUFFER> I'd hope you have that much on your working stack anyway  but I
          will put in a stack check, so at least it aborts "gracefully"

<[Wil] W.BADEN1>  STACK check not needed.  Just choose the smaller
          partition.  I will check offline.

<D.RUFFER> can that be done efficiently?...

<[Wil] W.BADEN1> Yes.

<D.RUFFER> I found the stack depth to be pretty dependent on the data
          content.

<[Wil] W.BADEN1> You have a lot of overhead already that u dont need.  The
          stack depth should not be data dependent.

<D.GOLDING> pass by address then?

<[Wil] W.BADEN1> that is, the max stack depth should not be data dependent.
          to don. of course.

<D.RUFFER> I will have to do some analysis...  I look forword to seeing
          what you can do to it too wil

<[Wil] W.BADEN1> I have a version in Zen that I did a couple of years ago,
          I'll exhume it.

<D.RUFFER> hum, sounds like the makings of a contest to me....

<D.GOLDING> or homework....

<D.RUFFER> who can come up with the fastest sort in one screen or less :-)

<[Wil] W.BADEN1> OSa cupl o'years ago I thot that I had found a ...  way to
          make Fastest KnowN Sort even faster.  This made me  hyper.  I was
          right -- but someone else had already done it.

<D.GOLDING> if you say one screen ot less, you will get unreadable,
          unuseable code, fewest words...better...

<D.RUFFER> yes, will have to set criteria based on words or some such
          measure

<D.GOLDING> better yet, fastest!

<D.RUFFER> I don't think I would go with just fastest...  you can do
          wonders by using a bunch of memory.

<D.GOLDING> if its there, use it!

<D.RUFFER> I think the most "elegent" solution would also be very
          conservative on memory

<D.GOLDING> I would like to see real world solutions that others can use,
          if you sort 10 i  10 items only but simple and fast, it would be
          futile...

<D.RUFFER> I agree, it would have to handle a pretty large set of data...
          I would probably also have to post the data set.  have we wound
          this topic down for tonight?

<[Wil] W.BADEN1> don -- is oc fig ready to discuss fast sorting in Forth?

<D.GOLDING> yes sir! how about at the advanced meeting?

<[Wil] W.BADEN1> Right by me.  Time to eat.

<D.RUFFER> what's os fig?

<[Wil] W.BADEN1> OSdon't we hav enuf with rtx chip?  orange county fig
          chapter(to which Don and I belong).

<D.GOLDING> I don't think we want to consume all of the time on it.

<D.RUFFER> ok, do enough people have interest in sorting techniques?

<D.GOLDING> I would say this is a very important topic...  Wil, that
          compression routine that used a binary tree was like a sorting al

<[Wil] W.BADEN1> No, some sorting algorithms ues a binary tree.

<D.GOLDING> that is what I meant

<[Wil] W.BADEN1> Dennis, why do u hav to carry 5 items on stack?

<D.RUFFER> well, the original did it all in variables...  I use the stack
          instead...  I tried to do some analysis...

<[Wil] W.BADEN1> No variable required in Forth.

<D.RUFFER> and had a couple more efficient methods...  I don't think the
          variables were needed either...  the original even made another
          stack <yuk>

<D.GOLDING> sounds like a canidate for local variables?

<[Wil] W.BADEN1> not really.

<D.RUFFER> but as I said, I had a bug until about 30 minutes before...
          this RTC, so I only had time to do the simplest...  version that
          was closest to the original...  that one worked!

<D.GOLDING> it would be nice with a problem like this one to use names,
          keep the items on the stack and let Forth keep track of it....

<D.RUFFER> you only need to deal with about 3 items on the top of the stack
          at any one given time  that was my liit anyway :-)  I do so hate
          having to use things like PICK and ROLL

<[Wil] W.BADEN1> i'm glad that there is still interest in sorting.

<D.RUFFER> I'm actually surprised that there was...  I just had a need for
          one, so tried to do it.

<D.GOLDING> it makes for Write Only Code...WOC...  Wil's favorite topic!

<[Wil] W.BADEN1> the grandchildren of quicksort put a stop to the  OSsearch
          for fastest sorting.

<P.THUTT> is here.

<D.GOLDING> What is OS??????

<[Wil] W.BADEN1> OSTo dennis--Don mean WOC not sorting.

<J.BROWN11> is here.

<D.RUFFER> I was wonder if he meant that my code was read only (oh no :-)

<[Wil] W.BADEN1> OSMy version of RedRyder (9.4) does that.

<D.RUFFER> hello Mr Thutt, and welcome back jack

<D.GOLDING> it looked pretty good...

<[Wil] W.BADEN1> Hi P.Thutt.

<P.THUTT> hello all I see Wil is still writing only?

<[Wil] W.BADEN1> Hi Jack.

<[Dennis] D.RUFFER> if we are going to give names again <grin>

<[Jack of BCFB] J.BROWN11> hi

<[Jack of BCFB] J.BROWN11> what's new

<[Dennis] D.RUFFER> we are just winding down on sort philosophy

<[Wil] W.BADEN1> Where is hillsboro?

<P.THUTT> hillsboro is in Oregon

<[Wil] W.BADEN1> OIC.

<[Dennis] D.RUFFER> anything else we want to talk about?

<P.THUTT> My first time on, however I talked with (listened to Wil at
          FORML87)

<[Wil] W.BADEN1> dennis--how do i find about updating RedRyder 9.4?

<[Dennis] D.RUFFER> go over to the Mac RT wil, they probably have it  or is
          it commercial ware?

<[Wil] W.BADEN1> P.Thutt -- help me place you.  To Dennis--commercial now
          and paid for.  but the company is no longer at the address I
          have.

<P.THUTT> I was interested in Novix, and liked your f83x phraser

<[Dennis] D.RUFFER> I'm sure they will have current information in the Mac
          RT, since most of them use it, or so they say.

<[Wil] W.BADEN1> oh yes, I remember. what's the P for?

<P.THUTT> paul

<[Wil] W.BADEN1> Dennis-what do I do? post a message in some topic?

<[Jack of BCFB] J.BROWN11> Paul  try /nam Paul

<P.THUTT>  I also chided someone about anagrams

<[Dennis] D.RUFFER> wil, try a SEArch in their BB

<[Wil] W.BADEN1> Paul-tell me more about anagrams.

<[Dennis] D.RUFFER> just go into their BB and type SEARCH /RedRider/ ALL
          or you might find the topic by just doing an INDEX

<[Wil] W.BADEN1> Dennis- I think I hav an early release with 2-3 bugs.

<[Paul] P.THUTT> ned w. bain i think  actually ned w. bail

<[Wil] W.BADEN1> Good one. I hadn't found that one yet.

<[Dennis] D.RUFFER> must be more neil baud jokes right?

<[Paul] P.THUTT> I have been stuck in the non-4th world at work recently,
          and wonder what's going on.

<[Wil] W.BADEN1> ewa blind    edw blain  ... about 30 so far.

<[Dennis] D.RUFFER> is anything NEW going on with Forth?

<D.GOLDING> The RTX2000 by Harris.

<[Dennis] D.RUFFER> no, the RTX4000 is new, but not out yet

<[Jack of BCFB] J.BROWN11> Dennis,  has Gary-s been on with any of the VP-
          Planner people yet?

<[Paul] P.THUTT> What is the 4000

<D.GOLDING> does the RTX-4000 have a 16 bit or 32 bit ASIC bus?

<[Dennis] D.RUFFER> he was going to do a practice session this weekend, but
          hadn't heard when  the 4000 is the 32 bit version  have no idea
          what size bus they have...  would immagine it has to be 32 bit.

<[Paul] P.THUTT> Is there float hardware for the rtx-4000?

<[Dennis] D.RUFFER> there isn't even hardware for the rtx4000 yet <grin>

<D.GOLDING> that's alot of pins: 32 data, 32 address, 32 ASIC plus
          interupts,etc.

<[Paul] P.THUTT> Ah yes, but with 4th engines the software exists before
          the hardware.

<[Dennis] D.RUFFER> yes, I'm sure you could put software floating point on
          it <grin>

<[Jack of BCFB] J.BROWN11> sure, send me one and I'll do it

<[Dennis] D.RUFFER> Mike {forget last name } from harris was on here
          earlier

<D.GOLDING> you could tie the Floating Point Processor of your choice on
          the ASIC bus...

<[Wil] W.BADEN1> gotta go.  zai jian.  17

<[Dennis] D.RUFFER> thanks for stopping by wil

<[Jack of BCFB] J.BROWN11> bye will

<D.GOLDING> Has anyone else played with a RTX 2000 here?

<[Wil] W.BADEN1> has left.

<[Paul] P.THUTT> Is anyone still concerned about mainstreaming 4th or is
          the world accepting it only as embedded

<[Dennis] D.RUFFER> define mainstreaming

<D.GOLDING> bye Wil, see ya at the advanced meeting...

<[Paul] P.THUTT> mainstreaming means acceptance by academics, computer
          press etc as development lang for "real products"

<[Dennis] D.RUFFER> we are trying to work on it, but it is a tough road

<[Paul] P.THUTT> Question arises from various "Is 4th dead?" articles

<D.GOLDING> The world wants bigger more computers for monster bug infested
          programs...

<[Dennis] D.RUFFER> join us on Thursday for a "real" Forth product

<[Paul] P.THUTT> What happens Thur

<[Dennis] D.RUFFER> VP PLanner will be our guests

<[Paul] P.THUTT> Whose 4th was their devl. environ.?

<[Dennis] D.RUFFER> I believe it was their own, but I think Jack knows
          more????

<D.GOLDING> 40 % of the code is machine code, I talked to Kent...

<[Paul] P.THUTT> What time Thur.

<[Dennis] D.RUFFER> 9:30 Eastern

<[Paul] P.THUTT> Hope to join then. bye for now

<[Dennis] D.RUFFER> see ya then

<D.GOLDING> bye Paul

<[Dennis] D.RUFFER> well folks, it's getting late for me also...

<[Paul] P.THUTT> has left.

<[Dennis] D.RUFFER> can we call this a night?

<[Jack of BCFB] J.BROWN11> I'm still here

<[Dennis] D.RUFFER> if I leave, you will loose the steno...

<[Jack of BCFB] J.BROWN11> what time is it there

<D.GOLDING> Yes where can I find info on Novix and RTX?

<[Dennis] D.RUFFER> but you are welcome to stay and enjoy...  it is now, at
          the tone 10:46:30pm

<D.GOLDING> 7:46 in California...

<[Jack of BCFB] J.BROWN11> same in BC Canada

<[Dennis] D.RUFFER> yes, well have fun...  see you guys later

=== End of Steno notes. ===