ForthNet@willett.UUCP (ForthNet articles from GEnie) (02/28/90)
Date: 02-26-90 (08:06) Number: 2973 (Echo) To: IAN GREEN Refer#: NONE From: JACK WOEHR Read: NO Subj: FORTH DATA STRUCTURES Status: PUBLIC MESSAGE > > How about declaring data structures like an array, or a queue. A >stack is automatic but I am unsure how to implement these other >elementary data structures. > Ian, the CREATE ... DOES> syntax is Forth's way of declaring a custom object. Anything created thusly pushes its data address to the stack when executed, then performs the associated DOES> code. : ARRAY ( #entries ---) \ this "defining" word will create arrays CREATE ALLOT \ allot { #entries } bytes DOES> ( index --- addr) \ which at runtime self-indexes + ; 20 ARRAY FOO ok 0 FOO . 1234 ok 1 FOO . 1235 ok ... etc. Note there are no limit checks in this construct! However, limit checks can be implemented. : CHECKARRAY ( #entries ---) CREATE DUP , ALLOT DOES> ( index --- addr | abort) OVER 0< ABORT" Negative array index!" 2DUP @ >= ABORT" Index out of range!" CELL + + ; 4 CHECKARRAY FOO ok 0 FOO . 1234 ok 1 FOO . 1235 ok -1 FOO . Negative array index! 4 FOO . Index out of range! For a more elaborate example, a self-indexing n-dimensional array, see DIMARRAY.ARC on most of these fine ForthNet stations. =jax= NET/Mail : RCFB Golden, CO (303) 278-0364 VESTA & Denver FIG for Forth! ----- This message came from GEnie via willett through a semi-automated process. Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'
ForthNet@willett.UUCP (ForthNet articles from GEnie) (03/07/90)
Date: 03-05-90 (09:14) Number: 2997 (Echo) To: DENNIS RUFFER Refer#: 2984 From: PETE KOZIAR Read: NO Subj: STACK USAGE Status: PUBLIC MESSAGE About all I can add to this is that, even though each word needs to put a return address on the return stack, the data manipulation words need to potentially do many accesses on the data stack. --- * Via Qwikmail 2.01 The Baltimore Sun ----- This message came from GEnie via willett through a semi-automated process. Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'
ForthNet@willett.UUCP (ForthNet articles from GEnie) (03/18/90)
Date: 03-13-90 (09:54) Number: 3035 (Echo)
To: ALL Refer#: NONE
From: JONAH THOMAS Read: (N/A)
Subj: B-TREES Status: PUBLIC MESSAGE
Some months ago Dave Weinstein put up a message about using TREE
structures, following a concept by Larry Forsley. There was a defining
word, TREE , which recursively made each word it defined to also be a
defining word which did the same thing TREE did. Each word got tagged
with the address of the word which defined it, which created a
tree-like hierarchy. Then he defined a set of words which looked at
the tags to trace out the 'lineage' of any particular word. It was
based on an idea by Larry Forsley. It looked neat, so I saved it.
When I needed it, I was a bit disappointed. It's a really nice
idea, but it doesn't actually do much. Each word in the tree is a
defining word. You use it to define 'branches', the later words in the
tree. Each of THEM is ALSO a defining word, which can define new
defining words. But none of the words can do anything except define
new defining words, so you need to fiddle with them to actually use
them for anything.
The obvious way to do it is to use a non-recursive defining word,
which would then be used something like:
0 subs car
car subs ford
car subs chevy
car subs mercedes
ford subs escort
chevy subs nova
nova subs 1987
Then CHILD could give its pfa, which contains the pfa of PARENT.
All nice and neat to use, just an extra word when you define a branch.
: SUBS CREATE , ;
: .NAMER BODY> >NAME .NAME ; ( For Pygmy, 3 - .ID )
: ANCESTRY BEGIN @ ?DUP WHILE DUP .NAMER REPEAT ;
---
* Via ProDoor 3.1
NET/Mail : The MATRIX (5 Nodes/1.2 Gig) Birmingham, AL (205) 323-2016
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'ForthNet@willett.UUCP (ForthNet articles from GEnie) (03/19/90)
Date: 03-15-90 (04:41) Number: 3038 (Echo)
To: ALL Refer#: NONE
From: JONAH THOMAS Read: (N/A)
Subj: B-TREES Status: PUBLIC MESSAGE
This is simpler than the recursive form to put together, but it
takes more typing to actually make the tree. What the recursive
defining words give you is the chance to take all those 'subs' out of
the tree set-up. But then you can only use the tree words with
specialized tree-reading words, in post-fix notation. I haven't
actually used it all enough to see whether this is a disadvantage.
The B-Tree was almost as easy:
: B-SIDE 2 + ;
: B-DOWN 4 + ;
: B-LINK-UP ( here addr | here 0 -- )
DUP IF B-DOWN DUP @ , ! ELSE , DROP THEN ;
: SUBS ( addr -- ) CREATE HERE SWAP DUP , B-LINK-UP 0 , ;
Three fields, the 1st looks back, the 2nd looks sideways, and the 3rd
looks forward. B-LINK-UP copies the parent's down-address into the
child's side-address, and the child's address into the parent's
down-address, so the 1st child defined (the last one in the list) at
each level has a 0 in its side-address.
VARIABLE B-LEVEL ( used for indenting while printing )
: SAME-LEVEL 2 + @ ;
: DOWN-ONE 4 + @ ;
: .TREE ( addr -- ) RECURSIVE CR B-LEVEL @ SPACES DUP .NAMER DUP
DOWN-ONE ?DUP IF 1 B-LEVEL +! .TREE -1 B-LEVEL +! THEN
SAME-LEVEL ?DUP IF .TREE THEN ;
A non-recursive form of .TREE was more complicated because it had to
keep track of what it was doing, and know when to quit. Recursion does
all that on the return stack automatically.
Recursive defining words ought to be really useful, but they don't
look that useful to me yet. Maybe with some special state variables, so
they're part-time recursive defining words. But it looks really
complicated.
---
* Via ProDoor 3.1
NET/Mail : The MATRIX (5 Nodes/1.2 Gig) Birmingham, AL (205) 323-2016
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: 'uunet!willett!dwp' or 'willett!dwp@gateway.sei.cmu.edu'ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/01/90)
Date: 05-29-90 (21:07) Number: 610 (Echo) To: ALL Refer#: NONE From: MICHAEL HAM Read: (N/A) Subj: EASTER IDENTIFICATION Status: PUBLIC MESSAGE Does anyone have a nifty algorithm that will give the month and day of Easter given the year? (I can't really believe that such a thing is possible given the definition of Easter.) If not, does anyone have a pointer to a table of the Easter dates from, say, 1900 to 2080? (If and when and I get this, I'll upload the table--at 360 bytes at the outside, it is probably smaller than the algorithm--and that's allowing a byte for month and a byte for day. And since we know Easter ranges from March 22 through April 25, we could do it in one byte: 022 thru 031 for the March dates, say, and 101 through 125 for the April dates, with the year used as the offset into the table (after subtracting 1900).) NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/01/90)
Date: 05-29-90 (20:22) Number: 3289 To: ALL Refer#: NONE From: MICHAEL HAM Read: HAS REPLIES Subj: EASTER LOCATION Status: PUBLIC MESSAGE Easter wanders around the calendar between March 22 and April 25; it is (in this part of the work) supposed to be the first Sunday after the first full moon after the vernal equinox--or is it the first new moon? At any rate, I have read that the algorithm actually used does not quite match the astronomical facts (for the sake of simplicity). And does anyone have the algorithm or know where I can see it? Alternately (and no doubt much simpler) can anyone point me to a table of Easter dates for the years 1900 through 2080? ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/01/90)
Date: 05-30-90 (20:17) Number: 3291 (Echo) To: MICHAEL HAM Refer#: 3289 From: MICHAEL HAM Read: NO Subj: EASTER LOCATION Status: PUBLIC MESSAGE Well, I have an answer about Easter Sunday: it falls on the first Sunday following the arbitrary Paschal Full Moon, which does not necessarily coincide with a real or astronomical full moon. The Paschal Full Moon is calculated by dividing the year by 19, and then using the remainder to index into the following table: 0 Apr 14 5 Apr 18 10 Mar 25 15 Mar 30 1 Apr 3 6 Apr 8 11 Apr 13 16 Apr 17 2 Apr 23 7 Mar 28 12 Apr 2 17 Apr 7 3 Apr 11 8 Apr 16 13 Mar 22 18 Mar 27 4 Mar 31 9 Apr 5 14 Apr 10 For the year 2000, for example, leaves a remainder of 5 when divided by 19. The table shows that the Paschal Full Moon then falls on April 18. In 2000, April 18 is a Tuesday, so thaM following Sunday (April 23) is Easter Sunday. NOTE: If the Pascal Full Moon falls on a Sunday, Easter is the following Sunday. The earliest Easter can Fall is March 23; the latest is April 25. Obviously the simplest thing will be to compute the Easter Sunday dates and then store them in a table: if I use a 0 prefix to denote March and a 1 prefix to denote April (so that 027 means March 27 and 115 means April 15), I can store each date in a single byte (with one bit left over for other uses), indexing into the table with the year normalized to the beginning of the table. Ash Wednesday, if you need that, is 40 days before Easter, not counting Sundays. Or 46 days, if you do count Sundays. ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/02/90)
Category 3, Topic 20
Message 92 Thu May 31, 1990
W.BADEN1 [Wil] at 21:07 PDT
Here is a vanilla algorithm for Easter.
( Easter is the first Sunday strictly after the ) ( Paschal Full Moon.)
: golden# ( year -- n) 19 MOD ( n) 1+ ;
: centuryterm ( year -- n)
100 / ( n) DUP >R 4 /
R@ 11 + 8 * 25 / + R> - ;
: paschalfullmoon ( year -- marchDay)
DUP golden# >R centuryterm ( n)
R@ 11 * + 30 MOD DUP 0= ( n f)
OVER 1 = ( n f f') R> 11 AND
OR ( n f) - ( n) 50 - NEGATE ;
: easter ( year -- mo da)
DUP >R paschalfullmoon ( n) 3 OVER ( n mar n)
R> wday - ( n) 7 + 31 /MOD ( d m) 3 + SWAP ;
Given month day year "wday" yields the day of week. (0 is Sunday.) Your system
may already have something for this. The following is for the Gregorian
calendar.
( For convenience the year begins with March 1.)
: yday ( month day anyYear -- dayOfYear year)
ROT ( day year month) 3 - DUP 0<
IF -1 +under 12 + THEN
306 * 5 + 10 / +under ;
: cday ( month day anyYear -- dayOfCentury century)
yday ( day year)
100 /MOD SWAP ( day century year)
1461 4 */ +under ;
: wday ( month day anyYear -- dayOfWeek)
cday ( day century) 4 MOD 2* 9 - - 0 7 UM/MOD DROP ;
Procedamus in Pace. Wil. (If you don't have it already : +under ROT + SWAP ;)
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.eduForthNet@willett.UUCP (ForthNet articles from GEnie) (06/04/90)
Date: 05-31-90 (21:01) Number: 3297 (Echo) To: MICHAEL HAM Refer#: 3291 From: JERRY SHIFRIN Read: 05-31-90 (23:13) Subj: EASTER LOCATION Status: PUBLIC MESSAGE MH>Well, I have an answer about Easter Sunday: it falls on the MH>first Sunday following the arbitrary Paschal Full Moon, which That must be when the Pascal programmers really go nutzoid! Just out of curiousity, why were you interested? A calendar program? --- ~ EZ 1.29 ~ ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/04/90)
Date: 05-31-90 (23:10) Number: 3298 (Echo) To: MICHAEL HAM Refer#: 3291 From: MICHAEL HAM Read: 05-31-90 (23:12) Subj: EASTER LOCATION Status: PUBLIC MESSAGE In the Paschal Full Moon table above, entry no. 2 is in error: it should be Mar 23 instead of Apr 23. I am uploading EASTER.ZIP, which contains a table of 176 Easter Sunday dates (1901-2076), along with two words to extract the date (given the year) and to print the date. The table works fine (used with UR/FORTH), but do NOT use INCLUDE to load the file--INCLUDE forces each byte to be at least 32, which zaps all the entries for March in ETABLE. So long as ETABLE is in the main file, it should work fine. ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/04/90)
Date: 05-31-90 (23:13) Number: 3299 (Echo) To: JERRY SHIFRIN Refer#: 3297 From: MICHAEL HAM Read: NO Subj: EASTER LOCATION Status: PUBLIC MESSAGE I am writing a program to help my wife schedule staff on the ward where she works (VA Hospital), and as an aid, I wanted a little pop-up calendar. Showing the Federal holidays was a relative snap, but Easter baffled me. (Easter not observed Federally, but if you're scheduling staff it's helpful to know.) ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/04/90)
Date: 06-01-90 (09:38) Number: 620 (Echo) To: ALL Refer#: NONE From: MICHAEL HAM Read: (N/A) Subj: EASTER TABLE Status: PUBLIC MESSAGE God must give us sleep so that we can see our mistakes while unconscious. I realized after posting EASTER.ZIP that it would be easy to make the table INCLUDE-proof by using BL to increment the values for March (which might fall under the INCLUDE-imposed minimum of 32). BL is about the right size, won't put March dates above 100, and is a constant so takes little room. EASTER2.ZIP is the INCLUDE-proof table. NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
rvp@cbnewsh.att.com (rob.v.phillips) (06/05/90)
In article <1044.UUL1.3#5129@willett.UUCP>, ForthNet@willett.UUCP (ForthNet articles from GEnie) writes: > Does anyone have a nifty algorithm that will give the month and day of > Easter given the year? ... I recall reading an algorithm, I believe in Apple or Atari Basic in an issue of "Creative Computing" around 1978-1980 that had this algorithm... Good Luck finding it! I think it was about 200 lines of code.
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/06/90)
Category 3, Topic 20
Message 97 Tue Jun 05, 1990
W.BADEN1 [Wil] at 20:39 PDT
At last Thursday's Fig Bar I promised two algorithms for calculating the date
of Easter. I gave the first the same night. I said that the other used
UM/MOD ten times. This is from T.H.O'Beirne's (That's O'BEIRNE) delightful
PUZZLES AND PARADOXES, Oxford Univ. Press, 1965, chapter titled "Ten Divisions
Lead to Easter."
My son programmed this for my Apple F83X some years ago, but rather than
excavating the code (I don't know where in California it is) or reproducing
O'Beirne's description, here is an executible program in ABC.
HOW TO DIVIDE dividend BY divisor GIVING quotient WITH remainder REMAINDER
PUT floor(dividend/divisor),dividend mod divisor IN quotient, remainder
HOW TO EASTER x:
\ Display month and day of Easter Sunday, Gregorian year x.
DIVIDE x BY 19 GIVING junk WITH a REMAINDER
DIVIDE x BY 100 GIVING b WITH c REMAINDER
DIVIDE b BY 4 GIVING d WITH f REMAINDER
DIVIDE 8*b + 13 BY 25 GIVING g WITH junk REMAINDER
DIVIDE 19*a + b - d - g + 15 BY 30 GIVING junk WITH h REMAINDER
DIVIDE a + 11*h BY 319 GIVING mu WITH junk REMAINDER
DIVIDE c BY 4 GIVING i WITH k REMAINDER
DIVIDE 2*f + 2*i - k - h + mu + 32 BY 7 GIVING junk WITH lambda REMAINDER
DIVIDE h - mu + lambda + 90 BY 25 GIVING n WITH junk REMAINDER
DIVIDE h - mu + lambda + n + 19 BY 32 GIVING junk WITH p REMAINDER
WRITE {[3]: "March"; [4]: "April"}[n], p
This seems wordy, but after you define a new keyword the system prompts you
with suggestions. E.g.,
When you start a line with "d", ABC echoes "D?ELETE ?" on top of your typing.
If you continue with "i" it suggests "DI?VIDE ? BY ? GIVING ? WITH ?
REMAINDER". You can write over the suggestion or use tab to copy up to the
next hole, given by "?". Notice that you do not have to shift case.
Here are some "how-to's" for stack manipulation.
HOW TO EMPTY stack
\ Also used to make stack.
PUT {} IN stack
HOW TO PUSH x ONTO stack
PUT x IN stack[#stack+1]
HOW TO RETURN top stack
RETURN stack[#stack]
HOW TO SWAP stack
PUT stack[#stack], stack[#stack-1] IN stack[#stack-1], stack[#stack]
HOW TO DROP stack
DELETE stack[#stack]
HOW TO DUP stack
PUSH top stack ONTO stack
HOW TO OVER stack
PUSH 1 th'on stack ONTO stack
HOW TO RETURN n th'on stack
RETURN stack[#stack - n]
Here is the clearest expression of The Sieve I've seen.
HOW TO SIEVE n
\ Sieve of Eratosthenes. (Not a benchmark.)
SHARE primes
PUT {2..n} IN primes
PUT 2 IN prime
PUT floor root n IN limit
WHILE prime <= limit:
PUT prime IN i
WHILE i < max primes:
PUT i + prime IN i
IF i in primes:
REMOVE i FROM primes
PUT prime min primes IN prime
WRITE "There are `#primes` primes <= `n`."
"SHARE primes" makes primes available globally.
Text files are treated as named tables with line number as the key and text as
the value of the key, so no I/O instructions are needed to manage them. There
is a built-in editor for all named data: numbers, strings, compounds, lists,
and tables.
Procedamus in pace. Wil.
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.eduForthNet@willett.UUCP (ForthNet articles from GEnie) (06/08/90)
Date: 06-01-90 (01:21) Number: 3323 (Echo)
To: MICHAEL HAM Refer#: 3291
From: KENNETH O'HESKIN Read: NO
Subj: EASTER LOCATION Status: PUBLIC MESSAGE
MH>Well, I have an answer about Easter Sunday: it falls on the
MH>first Sunday following the arbitrary Paschal Full Moon, which
MH>does not necessarily coincide with a real or astronomical full
Hmmm. Actually the calculation of Easter was a point of
bitter dogma and contention in the early Christian Church
(since nobody has or had a clue as to the historic birth/death
dates of Christ). Christmas was a straight appropriation of
of the pagan Winter Solstice (get all those unruly Goths
on side). The Celtic Church used a different algorithm for
calculating Easter which annoyed the Roman Church to no end,
and their rivalry (rather nasty at times) lasted a century.
Accuracy of calculation had nothing to do with it; it was
pure politics over who was going to run the store.
Maybe this thread should be reposted in the ANSI Conference
for the sober reflection and spiritual edification of the
Technical Committee ... :-)
---
~ EZ 1.26 ~ You never leave the bus -- Ken Keasy
NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.eduForthNet@willett.UUCP (ForthNet articles from GEnie) (06/08/90)
Date: 05-31-90 (18:42) Number: 628 (Echo) To: MICHAEL HAM Refer#: 610 From: JOHN SOMERVILLE Read: NO Subj: EASTER IDENTIFICATION Status: PUBLIC MESSAGE I remember reading somewhere that Greenwich uses the Testament to determine the date. No algorithm. As a point of interest, one of the Easters in the last tens years was observed one the wrong Sunday. . Strange, but true. regards j NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/08/90)
Date: 06-06-90 (10:48) Number: 3324 (Echo)
To: ALL Refer#: NONE
From: CHARLIE HITSELBERGER Read: (N/A)
Subj: PRIME NUMBERS Status: PUBLIC MESSAGE
I was just sort of messing around the other day, trying to write a
primes generator in Forth83. Here's what I came up with:
2VARIABLE NUMER 3. NUMER 2!
VARIABLE DENOM
: PRIMES ( -- ) CR
BEGIN 3 DENOM ! FALSE
BEGIN
DROP NUMER 2@ DENOM @
UM/MOD DROP DUP
IF
DENOM @ DUP M*
NUMER 2@ D>
2 DENOM +!
ELSE
TRUE
THEN
UNTIL
IF
NUMER 2@ 20 D.R
THEN
NUMER 2@ 2 M+ NUMER 2!
?TERMINAL
UNTIL
;
here's the BASIC equivalent:
10 N=3
20 D=3
30 IF N/D>INT(N/D) THEN 50
40 N=N+2:GOTO 20
50 IF D*D<N THEN D=D+2:GOTO 30
60 PRINT N,:GOTO 40
Is my method a pretty "elegant" approach? What does the standards
committee think that "1000000. 3 UM/MOD" is supposed to leave on the
stack? I'd leave a "1 65535" myself (the modulo and an overflow flag).
Of course, my implementation doesn't do that. It leaves invalid results
(the remainder is bigger than the divisor!) Also, what is a good way to
generate primes in Forth83? I wrote this as a benchmark, and sort of as
a way to get my feet wet again. Any comments would be appreciated.
Charlie Hitselberger
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.eduForthNet@willett.UUCP (ForthNet articles from GEnie) (06/10/90)
Date: 06-06-90 (07:31) Number: 3329 (Echo) To: KENNETH O'HESKIN Refer#: 3323 From: MICHAEL HAM Read: NO Subj: EASTER LOCATION Status: PUBLIC MESSAGE There is one group that is strongly advocating that Easter be made the second Sunday of April, which cuts the Gordian knot of the various algorithms, which can have the Eastern Church celebrating Easter one, two, three, four, or even five weeks after the Western Church. My search was simply for an algorithm that would give the Easter Sunday date for the US, as part of a staff scheduling program I am writing. NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/11/90)
Category 3, Topic 20 Message 104 Sat Jun 09, 1990 W.BADEN1 [Wil] at 14:26 PDT Oops. Examining last week's "paschalfullmoon" it is apparent that something is wrong with the next-to-last line: the code is incompatible with the stack comments. Restoring the lost ">" it should have been: : paschalfullmoon ( year -- marchDay) DUP golden# >R centuryterm ( n) R@ 11 * + 30 MOD DUP 0= ( n f) OVER 1 = ( n f f') R> 11 > AND OR ( n f) - ( n) 50 - NEGATE ; "easter" was also sick -- it yielded April 0 instead of March 31. : easter ( year -- mo da) DUP >R paschalfullmoon ( n) 3 OVER ( n mar n) R> wday - ( palmSunday) 7 + 3 SWAP ( mo da) DUP 31 > IF 1 +under 31 - THEN ; I'm very sorry and I won't do it anymore. ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/11/90)
Date: 06-09-90 (14:07) Number: 3337 (Echo)
To: W.BADEN1 [WIL] Refer#: 3305
From: MICHAEL HAM Read: NO
Subj: EASTER AGAIN Status: PUBLIC MESSAGE
Wil, I took a slightly different approach, though I am intrigued by the
way you did it. Here's mine:
My approach to the Easter problem was to construct a table of the
values--I believed that this approach offered advantages in speed
and probably in memory room as well. I was wrong about the
memory: a useful set of words using the computational approach
shown below is 142 bytes versus 224 for the table apprach, but
the speed advantage is all to the table: 4.12 seconds for 10,000
Easter computations versus .49 seconds for 10,000 Easter lookups,
with the result biased by the difference in speed for 3DROP
versus 2DROP--not enough to account for the difference.
The table of dates is given below, along with two example words:
one to fetch the Easter date and one to display it. The table is
normalized to 1901 (i.e., entry number 0 (lower-case k, as you
see) is the entry for 1901), with the Easter Sunday date for
each year (through 2076) stored in a single byte.
Initially I tried using the date of the month with 0 prefixed for
dates in March and 1 for dates in April (for example, 028 for 28
March; 120 for 20 April). However, UR/FORTH's INCLUDE (used to
load files) forces bytes to a minimum value of 32, which messed
up the March dates. I therefore added BL to each March date to
make it immune to alteration by INCLUDE.
The UR/FORTH word ," stores into memory the (uncounted) string
that follows, up to but not including the delimiting ". In the
screen dump below, the lines break at the end of each screen
line, but the table is one contiguous string of 176 (non-blank)
characters.
CREATE ETABLE \ INCLUDE-proof table of Easter Sundays 1901-2076
," k>pg{s?wo;tk7ph{l?xh;texphul?xi;teyp<um8qi}meyj<um9qivneyj=uf
zr=vn:rj=ofzr>vn:sjwogzk>wg:s?wogtk>ph{s?xo;tl7ph|l?xi;teyphum?x
i<teyq<um9qi}neyj=um9rivnfyj=vfzr>vn:sj=ogzr>wn:skw"
: @EASTER ( yr - da mon | flag ) DUP 1901 < OVER 2076 > OR
IF ." The year " U. ." is out of range for this table. " 0
ELSE 1901 - ETABLE + C@ DUP 100 >
IF 100 - 4 ( Apr ) ELSE BL - 3 ( Mar ) THEN THEN ;
: .EASTER ( yr - ) DUP @EASTER ?DUP
IF 4 = IF ." April " ELSE ." March " THEN 0 .R ." , " .
ELSE DROP ( yr ) THEN ;
The table was constructed using the following words, which are
merely scaffolding to be discarded once the tiny edifice of the
table is complete.
3 CONSTANT MAR \ for readability
4 CONSTANT APR
: | ( n1 n2 - ) , , ; \ for readability
CREATE 'PFM 14 APR | 3 APR | 23 MAR | 11 APR | 31 MAR |
( Paschal ) 18 APR | 8 APR | 28 MAR | 16 APR | 5 APR |
( Full Moon ) 25 MAR | 13 APR | 2 APR | 22 MAR | 10 APR |
( table ) 30 MAR | 17 APR | 7 APR | 27 MAR |
: CK-PFM PRINTER 19 0 DO I 4 * 'PFM + 2@ 3 =
IF ." March" ELSE ." April" THEN 3 .R CR LOOP
12 EMIT CONSOLE ;
It is very handy for any table to have a way to dump the entries
to hard copy in readable form to simplify proofreading.
: UMOD ( u1 u2 - urem ) 0 SWAP UM/MOD DROP ;
\ 0 SWAP converts the single u1 to a double, for UM/MOD
: PFM ( yr - day mon ) \ yr in, day/mon of Paschal Full Moon out
19 MOD 4 * 'PFM + 2@ ;
: EASTER ( yr - day mon yr ) \ given year, gives date of Easter
DUP PFM ROT >CENT DUP 7 UMOD 7 SWAP - + CENT> ;
The word >CENT in my calendar routine converts day mon yr to a
century date, an offset from 1/1/1900 (accurate from 3/1/1900--
1900 is not a leap year, but the simple algorithm I use is
ignorant of the subtleties of the Gregorian scheme). CENT>
converts a century date to day month year. (George Shaw pointed
out to me that you can check a date's validity by converting it
to century format and back: the date is valid if and only if the
initial and final values match.)
Using 7 UMOD on a century date gives the day of the week, by
happy chance Sunday being 0. By subtracting the weekday value
(which thus happens to be the offset from Sunday) from 7 and
adding it to the date, we get the next Sunday.
CREATE EASTERS 176 ALLOT
: FILL-TABLE 176 0 DO 1901 I + EASTER DROP 4 =
IF 100 + ELSE BL + THEN EASTERS I + C! LOOP ;
Once the table was created, it was easy to store it into the
appropriate block with CMOVE.
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.eduForthNet@willett.UUCP (ForthNet articles from GEnie) (06/14/90)
Date: 06-11-90 (06:58) Number: 639 (Echo) To: JOHN SOMERVILLE Refer#: 638 From: MICHAEL HAM Read: NO Subj: PROJECT MANAGEMENT Status: PUBLIC MESSAGE I would think that the best way to load a set of small files is a series of INCLUDE statements in the main file: INCLUDE ABC.EFG will load the file ABC.EFG. Note that the extension must be included in the filename. Note also that INCLUDE will force bytes to a minimum value of 32 (ASCII blank), so if you have data tables, for example, with bytes lower in value, beware. NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/14/90)
Date: 06-11-90 (08:56) Number: 640 (Echo) To: JOHN SOMERVILLE Refer#: 638 From: RAY DUNCAN Read: NO Subj: PROJECT MANAGEMENT Status: PUBLIC MESSAGE The best solution is to have one "master" file that just contains a bunch of INCLUDE statements for other files. We use this technique quite a bit here at LMI. NET/Mail : LMI Forth Board, Los Angeles, CA (213) 306-3530 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/16/90)
Date: 06-13-90 (20:37) Number: 644 (Echo) To: MICHAEL HAM Refer#: 639 From: JOHN SOMERVILLE Read: NO Subj: PROJECT MANAGEMENT Status: PUBLIC MESSAGE Thanks that will probably do. Most tables that I use I could copy to the main file. . I have been racking my brains trying to recall where I read about the Easter holiday reckoning at Greenwich. I was astonished when I read it, but assumed the information was ( about the Testament was correct, maybe it is.) Paschal comes from what. By the way I have enjoyed your articles a lot. regards j NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.UUCP (ForthNet articles from GEnie) (06/16/90)
Date: 06-13-90 (20:42) Number: 645 (Echo) To: RAY DUNCAN Refer#: 640 From: JOHN SOMERVILLE Read: NO Subj: PROJECT MANAGEMENT Status: PUBLIC MESSAGE Thanks Michael Ham suggested the same thing. . I received 1.1 yesterday. I have started to read the manual, and understand why you recommend getting it. Note page I - 24. I tried the example on J - 541 and it didn't link is that because I am using LINK 3.05?. Also Map0.obj is non-existent; I presume MAP.OBJ is its replacement. regards j NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886 ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or willett!dwp@hobbes.cert.sei.cmu.edu
ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie) (09/12/90)
Date: 09-08-90 (10:53) Number: 484 (Echo)
To: ALL Refer#: NONE
From: DARRYL BIECH Read: (N/A)
Subj: CONTROL STRUCTURES Status: PUBLIC MESSAGE
Being new to FORTH, but experienced in several other languages, a couple
of things puzzle me about the FORTH strings and control structures
proposals/debates:
>
1.) What is wrong with using the three conventional looping constructs
DO FOR, DO WHILE, and DO UNTIL, and combining them with IF THEN
(ELSE) to create the more complicated control structures that I
noticed at the X3J14 Vancouver Conference. Using iterative and
"decide before the body of the loop" and "decide after the body of
the loop" constructs are enough for any language.
(I am referring to technical comment C89-279.1)
2.) Why not have counted strings that are optionally null-terminated?
NET/Mail : British Columbia Forth Board - Burnaby BC - (604)434-5886
-----
This message came from GEnie via willett through a semi-automated process.
Report problems to: uunet!willett!dwp or dwp@willett.pgh.pa.usForthNet@willett.pgh.pa.us (ForthNet articles from GEnie) (09/13/90)
Category 3, Topic 20 Message 115 Wed Sep 12, 1990 D.RUFFER [Dennis] at 00:29 EDT Re: DARRYL BIECH > 1.) What is wrong with using the three conventional looping > constructs DO FOR, DO WHILE, and DO UNTIL, and combining them with > IF THEN (ELSE) to create the more complicated control structures I don't rmember the comment you refer to, but what we now have, IMHO is very close you what you are suggesting. Forth uses different commands, so here are the equivalents: Traditional Forth DO FOR -> DO LOOP DO WHILE -> BEGIN WHILE REPEAT DO UNTIL -> BEGIN UNTIL IF THEN (ELSE) -> IF (ELSE) THEN There seems to be a lot of confusion about how these individual operators can be combined, and my understanding is that the current BASIS does not even reflect what it should in this regard (i.e. what has already been passed by the TC). Perhaps only experimentation with a compliant system will tell the real answer, but I have been assured that the following work: BEGIN WHILE UNTIL THEN Thus creating a loop that has two exit points. I made a proposal very early in the proccess for a similar construction. It was rejected in favor of the more powerful generalizations that we have now. I think that is what you are looking for also. > 2.) Why not have counted strings that are optionally null- > terminated? Tradition? It would be a counted string if it was null-terminated. Why not terminate strings with a $? What about a user selectable character? Where would it all end? :-) DaR ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or dwp@willett.pgh.pa.us
cwpjr@cbnewse.att.com (clyde.w.jr.phillips) (09/13/90)
In article <1716.UUL1.3#5129@willett.pgh.pa.us>, ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie) writes: > > 2.) Why not have counted strings that are optionally null-terminated? I agree we should have it but ynot is because it's bowing to C peer pressure. I think having to "scan" a string to determine it's length is bad ( its Bahhd ). We have a little overhead for sureties sake. However if we make a null terminator manditory AND "counted" a little more overhead ( stop on nulls ) might add a "good" flavor to our strings ( I would have said spaghetti but you never know what type of person is reading this! ) I'll get around to assessing this sooner or later. Clyde
a684@mindlink.UUCP (Nick Janow) (09/14/90)
In article <1716.UUL1.3#5129@willett.pgh.pa.us>, ForthNet@willett.pgh.pa.us (ForthNet articles from GEnie) writes: > 2.) Why not have counted strings that are optionally null-terminated? I wouldn't put them in ANS FORTH, since null-termination of a counted string would be redundant information. It's something that isn't used all that often. The nice thing about FORTH is that if your application needs null-terminated counted strings, uncounted packed strings or uncounted strings terminated with Japanese Haikum (who knows what someone might need?), you have the tools to add words to handle them. Go FORTH and write your null-terminated counted string package! :-)
wmb@MITCH.ENG.SUN.COM (09/14/90)
> 2.) Why not have counted strings that are optionally null-terminated?
In my system, counted strings are *always* null-terminated. The null
is not included in the count. This makes it easy to pass strings to C.
HOWEVER, I do not recommend this! I recommend that counted strings be
*eliminated entirely*. Forth should only have one visible string
representation (currently it has 4), and that representation should be
"adr len" on the stack. The "adr len" representation is by far the
most flexible, and does not suffer from the fundamental problems of
counted strings or null-terminated strings.
Those problems are:
1) Counted strings have an inherent length limitation.
2) You can't just point to some memory and say it's a counted string or
a null-terminated string; instead you have to copy it somewhere so you
can insert either the count byte or the null byte.
3) Null-terminated strings cannot represent byte arrays containing the
null byte.
4) Extraction of substrings of counted strings and null-terminated strings
requires copying.
The only down side of "adr len" strings is the stack manipulations required,
and that turns out not to be too bad once you get used to using "2dup"
and "2swap".
Over time, I am weeding out all externally-visible uses of counted strings
in my system.
Mitch Bradley, wmb@Eng.Sun.COMForthNet@willett.pgh.pa.us (ForthNet articles from GEnie) (09/15/90)
Date: 09-13-90 (10:57) Number: 3770 (Echo) To: CHARLIE HITSELBERGER Refer#: 3746 From: GENE LEFAVE Read: NO Subj: IS THIS COMMONLY DONE? Status: PUBLIC MESSAGE CH>what I wound up with. My question is, is the " 0 BEGIN DROP " a CH>commonly used thing for loops that leave a value on the stack upon exit? I woundn't know if it's commonly done, but I use it all the time. 8-) --- ~ EZ-Reader 1.13 ~ ----- This message came from GEnie via willett through a semi-automated process. Report problems to: uunet!willett!dwp or dwp@willett.pgh.pa.us
andrew@idacom.uucp (Andrew Scott) (09/20/90)
Mitch Bradley writes: > Forth should only have one visible string > representation (currently it has 4), and that representation should be > "adr len" on the stack. How do I get "adr len" on the stack? The operation is defined for strings defined with " , but what about strings in memory? I can only get the address of a string buffer defined with CREATE, for example. I must use COUNT, or perhaps STRLEN for a null terminated string, to get the length. I do not want to have to declare a seperate VARIABLE for the length of each string buffer I use. -- Andrew Scott | mail: andrew@idacom.uucp | - or - ..!uunet!ubc-cs!alberta!idacom!andrew
wmb@MITCH.ENG.SUN.COM (09/22/90)
> How do I get "adr len" on the stack? The operation is defined for strings > defined with " , but what about strings in memory? I can only get the > address of a string buffer defined with CREATE, for example. I must use > COUNT, or perhaps STRLEN for a null terminated string, to get the length. > I do not want to have to declare a seperate VARIABLE for the length of > each string buffer I use. : string: \ name ( max-size -- ) create 0 , allot does> ( pfa -- adr len ) dup cell+ swap @ ; The operators for manipulating strings should be as general as possible, and independent of how those strings happen to be encoded when stored in memory. One might reasonably choose different memory encodings for different applications, but the stack representation should always be the same, so that we could build upon a single set of string operators, rather than having the piecemeal set of operators, split among several different string representations, that we have now. Mitch Bradley, wmb@Eng.Sun.COM
shri@ncst.ernet.in (H.Shrikumar) (10/02/90)
In article <9009220053.AA13718@ucbvax.Berkeley.EDU> wmb%MITCH.ENG.SUN.COM@SCFVM.GSFC.NASA.GOV replies to ... >> How do I get "adr len" on the stack? The operation is defined for strings >> defined with " , but what about strings in memory? I can only get the > >The operators for manipulating strings should be as general as possible, >and independent of how those strings happen to be encoded when stored >in memory. Yup. Also could overload the same ideogram for same action on strings of differnt types, the correct word being selected by the value of a variable called TYPE, (much as numbers are interpretted according to current value of BASE). (* I saw a feature with similar intensions first in Mikael Patels xforth.) Also, another thing I am trying out in myForth is that all operators which take a pointer, should return a pointer. The idea is ... say a word that takes a string and replaces "$1" by "gus" ("fun$1prat" becomes "fungusprat") would take a pointer to the string, do its job and return a pointer to the new string. The retuned pointer may or may not be the same, the program must not make assumptions about it (sorry hackers :-). This is not too difficult to implement, only a little inefficient, and gives a clean semantics. (IMHO, worth it.) For eg. the string could be in ROM, and when somebody tried to modify it, the word could (if so designed) copy it into RAM "transparently", change it and return a new pointer. This could also become the basis for several other features (memory de-fragmentation etc.) -- shrikumar ( shri@ncst.in )
dwp@willett.pgh.pa.us (Doug Philips) (10/04/90)
In <978@shakti.ncst.ernet.in>, shri@ncst.ernet.in (H.Shrikumar) writes: > Also, another thing I am trying out in myForth is that all operators > which take a pointer, should return a pointer. > > The idea is ... say a word that takes a string and replaces "$1" by > "gus" ("fun$1prat" becomes "fungusprat") would take a pointer to the > string, do its job and return a pointer to the new string. > > The retuned pointer may or may not be the same, the program must not > make assumptions about it (sorry hackers :-). > > This is not too difficult to implement, only a little inefficient, > and gives a clean semantics. (IMHO, worth it.) The only objection I see to that is that it makes data structures that share data by pointer a pain, because anytime something happens to an object, there can be arbitrary many pointers that need to be updated. For that reason, and others, most systems without Virtual Memory (where an address can be "mapped" to any other arbitrary address at will and at run-time) go to another level of indirection, often called 'handles'. I can see how what you are proposing could have handles nicely layered on top of it. -Doug --- Preferred: dwp@willett.pgh.pa.us Ok: {pitt,sei,uunet}!willett!dwp