***** *** * * **** ***** ***** *** * * * ** * * * * * * * * * * * * * * * * * **** * **** * * * * * * *** *** * * * * * * * * * * * * * ** * * * * * * **** ***** * * **** * ****** *** Volume 3 Number 5 48/39/38 May 1978 Newsletter of the SR-52 Users Club published at 9459 Taylorsville Road Dayton, OH 45424 - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - An Analysis and Comparison of 3 Calendar Programs (59/PC-100A) Both Jared Weinberger (221) and Lou Cargile (625) have written programs to generate and print calendars, using the ML-20 CROM program, and outputting in a standard month-by-month weekday column format. This is the sort of problem where neither its complexity nor the best approaches to its solution are apt to become fully apparent until at least one approach has been tried. And so it wasn't until I examined their programs in detail that I appreciated the degree of the challenge and was motivated to try to find a better approach. Both programs require 3 card-sides for recording, but neither uses all the recordably available memory, and since it is unlikely that a comparable program could be made to fit on a single mag card, the payoff in optimization is in the reduction of run time. With the advantage of having Jared's and Lou's programs in hand I wrote a third one that runs in just three fourths the time required by either of the others (26 minutes versus 34-35 to print 12 months). The analysis and comparisons which follow are intended to be informa- tive and constructive, and I hope it goes without saying that it is not my intention to impugn either Jared's or Lou's programming abilities. In the same spirit, I welcome better approaches from others who now have all 3 programs to start with, and it is my hope that we will all learn something from this exercise. In addition to the usual user instructions and program listings, I am including high-level algorithms, which may make it easier to see how each program works. The most critical function to try to optimize is the generation of the print code required to print the successive integers 1,2,...31. Jared's approach to this requirement is the slow- est, partly because he needed a routine to translate 4-digit year numbers besides the day numbers. Lou dodged this requirement by let- ting the printer print the year as the displayed value, with the month tag-printed in the same line via Op 4 Op 6. But I expect that a routine optimized to only translate integers in the 1-31 range into print code would still run slower than a print-code counter. Lou devised just such a counter, and this difference appears to give his program a slight speed advantage. But both Jared and Lou use a number of x-t comparisons where they could have used Dszs (which are faster) for branch control in their print code generators, and Lou uses 4 flags, 3 of which require repeated setting and resetting, as well as testing. While inefficient approaches to meeting requirements posed only once per printed month aren't so critical as the once-per-day ones, the - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
extra execution time can add up. Both Jared and Lou constructed independent sequences to calculate the number of days in a month, since it apparently didn't occur to them that given the first day of 2 suc- cessive months, ML-20 returns the number of days in the first month. There is also some time lost putting month, day and year information into the required ML-20 input format, instead of directly storing the separate values in Reg 1, 2, and 3, and processing via SBR 086 or 177 (see V2N11p1 and Fred Fish's ML Survival Kit p 20-1). So try all 3 programs, look them over, and see if you can write a faster one (one which runs say, under 2 minutes per month). Incident- ly, although the ML manual doesn't say so, 9999 appears to be the upper limit on the year, unless inputs are stored separately. But at some point the 366-day year every 4 years rule won't hold. TI-59/PC-100A Program: Calendar Printer Jared Weinberger (221) Algorithm: 1. Print the month header. 2. Call ML-20 D with 1st day of specified month and year to get day of week; convert the returned 1,2,...6,0 to 1,2,...7. 3. Determine number of days in month: For Feb: Call ML-20 A with Feb 28th, and ML-20 B with March 1st to get the number of days between Feb 28th and March 1st; use this result to get the number of days in Feb. For other months: Test for April, June, Sept, Nov to get the number of days. 4. Fill a cleared block of 7 registers with consecutive base 10 inte- gers beginning with 1 in a position determined by the corresponding day of the week. 5. Convert the stored integers to print code via a translator; pack print buffers, and print a line of days. 6. Clear the integer block, fill with the next group of integers, and go to step 5. 7. Repeat steps 5 and 6 until all days have been printed. 8. For whole year: Increment month, and repeat steps 1-8 until all months have been printed. User Instructions: For one month: key month, press A; key year, press B; press D. For whole year: Key year, press B; press E. Program Listing: 000: LA S29 rtn LB S30 rtn CP x=t 036 ÷ log Int S8 Op28 INV log + 024: x:t 100 Prd 21 8 x≥t 036 2 + 1 - Int SUM21 = X 10 Dsz8 023 CLR 050: Exc21 x:t 1 x=t 059 x:t rtn 0 rtn R1 SBR 010 X 1 EE 6 = INV EE 073: S28 R2 SBR 010 + R28 = Op1 R3 SBR 010 X 1 EE 4 = INV EE S28 R4 ÷ 103: 10 = Int SBR 010 + R28 = Op2 R4 ÷ 10 = INV Int X 10 = SBR 010 S0 133: R4 CP x=t 148 R0 INV x=t 148 + 1 = EE 8 S28 R5 SBR 010 X 100 + 162: R28 = Op3 R6 SBR 010 X 1 EE 6 = INV EE S28 R7 SBR 010 + R28 = Op4 192: Op5 rtn LD Op0 R29 + 34 = S28 R*28 Op1 R30 SBR 010 Op4 Op5 Adv R31 223: Op1 R32 Op2 R33 Op3 R34 Op4 Op5 Adv R29 X 2 INV log + R30 ÷ 4 INV 252: log + 1 = Pgm20 D CP INV x=t 265 7 S10 R29 x:t 2 INV x=t 316 R30 277: ÷ 4 INV log = S0 + 228 = Pgm20 A R0 + 301 = Pgm 20 B Pgm20 C x:t 306: 1 x=t 311 + 28 GTO 340 R29 x:t 4 x=t 338 6 x=t 338 9 x=t 338 11 333: x=t 338 1 + 30 + 1 = S11 Pgm 1 SBR CLR 8 x:t 1 S13 R13 S*10 1 359: SUM10 SUM13 R10 INV x=t 354 SBR 061 0 S7 Pgm 1 SBR CLR 1 S12 R11 384: x:t R13 x=t 413 8 x:t R13 S*12 1 SUM12 SUM13 R12 INV x=t 382 SBR 408: 061 GTO 372 SBR 061 Adv Adv rtn LE 0 S29 1 SUM29 SBR 197 12 x:t 433: R29 INV x=t 424 Adv Adv rtn 52-NOTES V3N5p2
Prestored Data: 31: 3641003032 37410043 1700372300 2135003613 2513314000 2117144000 37: 3013351523 1333352427 3013450000 2541311700 2541274500 1341224000 43: 3617333740 3215374000 3132424000 1617154000 - - - - - - - TI-59/PC-100A Program Calendar Printer Lou Cargile (625) Algorithm: 1. Determine whether to print 1 month or balance of specified year. 2. Call ML-20 D with 1st day of the specified month and year to get day of week for 1st day; convert returned 1,2,...6,0 to 1,2,...7. 3. Check for leap year, Feb, odd or even month, July, Aug, to get number of days per month. 4. Print month header. 5. Use the calculated day of week to determine where to begin filling print buffers from a quasi base seven counter; print filled buffers and refill from the counter until all days have been printed. 6. For more than one month: Repeat steps 2-5, incrementing the month until all months for the specified year have been printed. User Instructions: For one month: Key MM.YYYY, press B; for balance of year: Key MM.YYYY, press A. Program Listing: 000: 12 x:t R8 INV x≥t 010 R/S 1 SUM6 GTO 026 LA INV Stflg4 INV Stflg 023: 0 S6 R6 S0 Int S9 INV SUM0 X 100 + 1 = SUM0 R6 INV Int X 10000 = 056: S10 R0 Pgm20 D S07 0 x:t R7 INV x=t 076 7 SUM7 R10 ÷ 4 = INV Int 083: INV x=t 089 Stflg0 31 S13 0 x:t R8 ÷ 2 = INV Int INV x=t 134 R8 108: x:t 2 INV x=t 129 29 S13 INV Ifflg0 145 1 SUM13 GTO 145 7 x:t 131: GTO 138 R8 x:t 7 INV x≥t 145 1 SUM13 R8 + 47 = S9 Op0 R*9 Op4 159: R10 Op6 SBR368 R46 Op1 R45 Op2 R44 Op3 R43 Op4 Op5 SBR369 INV 188: Stflg1 INV Stflg3 1 S00 0 S14 S16 5 S15 INV Stflg2 R14 SBR289 R14 214: SUM16 SBR236 Ifflg2 204 SBR236 Ifflg2 204 SBR236 GTO 204 Dzz15 238: 275 R16 Int Op*0 Op20 R0 x:t 5 INV x=t 266 1 S0 Op5 Stflg2 Ifflg3 264: 281 R16 INV Int S16 5 S15 100 Prd16 rtn Adv INV Ifflg4 288 R/S 288: RST Dsz7 338 Ifflg1 304 .01 SUM14 Stflf1 .01 SUM14 INV Dsz13 332 314: R14 INV Int x:t .08 x=t 339 .13 x=t 345 rtn 0 S14 Stflg3 rtn rtn 339: .02 SUM14 rtn .88 SUM14 R14 Int x:t 1 x=t 359 rtn SUM14 rtn LB 364: Stflg4 GTO 021 R47 Op1 Op2 Op3 Op4 Op5 rtn Prestored Data: 43: 21000036 4300003700 370000 36000030 2020202020 25133100 21171400 50: 30133500 13333500 31034500 25413117 25412745 13412200 36173337 57: 32153700 31324200 16171500 - - - - - - - - - - - TI-59/PC-100A Program: Calendar Printer Ed Algorithm: 1. Check for ML module presence; alert user with flashing display if absent. 2. Determine user's choice: 1 month, balance of start year, or print from start month and year until end of specified year. 3. Call ML-20 SBR 086 with 1st day of specified month and year to get corresponding absolute day, and ML-20 SBR 177 to get corresponding day of week. Call ML-20 SBR 086 with 1st day of next month and calculate difference between returned absolute day value, and first one to get number of days in specified month. 4. Print month headers. 52-NOTES V3N5p3
5. Use calculated day of week to initialize a general purpose print-
buffer-packer routine to print first line of days, followed by suc-
cessive lines until all days have been printed. Use a quasi base
seven counter to supply the print-buffer-packer routine with code
for successive integers.
6. For more months, as determined by step 2, cycle months, and incre-
ment years as required; repeat steps 3-5 until done.
User Instructions:
1. Initialize: Press A, see module confirmation printed; flashed dis-
play indicated wrong module.
2. Key start year, press R/S, key start month, press R/S.
3. For 1 month: Press B;
For balance of year: Press C; for specified period: Key end year,
press D.
Program Listing:
000: LC' 5 S17 1 EE 8 S19 CLR R21 Op*20 0 S21 1 SUM20 Dsz18 037 1 S20
028: Op5 Op0 Stflg0 4 S18 rtn LE' Op28 INV Dsz11 053 INV Dsz12 066 rtn
053: 1 S8 Op29 10 S11 8 S12 rtn 2 SUM08 9 S12 rtn LD' INV Stflg0 CLR 1
080: x:t E' R9 x=t 093 X R19 = SUM21 100 INV Prd19 INV Dsz17 145 R8 X
107: R19 = SUM21 100 INV Prd19 Dsz15 128 0 S18 C' Adv rtn INV Dsz17 149
133: 100 INV Prd19 Dsz17 D' C' GTO D' C' GTO 104 C' Ifflg0 D' GTO 133
156: LA' Op00 1 S9 S09 10 S11 8 S12 5 S17 4 S18 1 EE 8 S19 1 S20 0 S21
189: R16 x:t 1 x=t 232 2 x=t 234 3 x=t 244 4 x=t 260 5 x=t 282 6 x=t
214: 274 2 S17 1 S18 100 S19 4 S20 D' rtn D' rtn 2 S17 100 S19 D' rtn
244: 4 S17 3 S18 1 EE 6 S19 2 S20 D' rtn 1 S17 3 S18 1 S19 2 S20 D'
273: rtn 1 S18 4 S20 D' rtn 3 S17 2 S18 1 EE 4 S19 3 S20 D' rtn LA 1
301: x:t Pgm1 SBR Write x=t 312 Op55 R/S Adv Adv Adv R/S S13 R/S S14
321: R/S LB R14 S01 1 S2 R13 S3 Pgm20 SBR086 S10 Pgm20 SBR177 S16 R14
351: + 1 = S1 R13 S3 Pgm20 SBR086 - R10 = S15 Op0 R14 + 47 = S20 R*20
383: Op4 R13 Op6 R47 Op1 R46 Op2 R45 Op3 R44 Op4 Op5 A' rtn LC 13 -
414: R14 = S22 B 1 SUM14 Dsz22 419 CLR rtn LD - R13 + 1 = S23 C 1 SUM13
443: S14 Dsz23 439 Adv Adv R/S
Prestored Data:
44: 2135003613 1700372300 37410043 3641003932 25133100 211714 301335
51: 133335 310345 25413117 25412745 134122 36173337 321537 313242
59: 161715
- - - - - - - - -
Program/Routine Copyright
Legal and legislative bodies are recognizing that it is difficult
to apply copyright law (even the new one) to computer-readable works,
so it is not surprising that we laymen are confused as to what is and
isn't allowed. Ken Widelitz, a lawyer who writes the Legal/Business
Forum in Kilobaud magazine, addresses this topic (Nov 77 p14 and Apr 78
p6) in some detail. An important concept which Widelitz touches on is
the so-called Fair Use Doctrine, which effectively says that the less
you copy, and the farther removed your use is from competing in the mar-
ketplace with the original, the surer you can be that you are acting
within the law. An important widely accepted legal interpretation says
that ideas themselves are not copyrightable; the expression of them is.
Then there is the question of "derivative" works. Widelitz notes that
there is no clear-cut dividing line between vary basic techniques (such
a commonly used sorting routines) which are generally held to be uncopy-
rightable, and more specialized ones which are, and recommends that
52-NOTES V3N5p4
users seek legal advice for important specific cases.
Interested members will find Widelitz' 2 articles both informative
and thought provoking, and may find helpful a new National Bureau of
Standards publication (#500-17, $4.00 USGPO Wash DC 20402): "Copyright
In Computer-Readable Works: Policy Impacts of Technological Change".
Dave Johnston (5) and Lee Eastman (713) wonder if translations of
programs for different machines are protected by original copyrights.
I suggest applying the Fair Use Doctrine, and if it appears that your
intentions might conflict significantly with the originator's interests,
get legal advice. Members who have been involved in computer-readable
copyright litigation are invited to share their experiences.
52-NOTES Subject Index
Many of you have suggested that it would be helpful to have a
topical index, and I agree, except perhaps as to how it should be gen-
erated and maintained. To a large extent, the choice of alphabetically
listed names to categorize published material is subjective, and I sus-
pect that the best index is generated by the user himself, and maintained
as a continuously updated card file. Getting started is a bit of a
chore, but refamiliarizes the user with the published material, as well
as providing him with a relevant reference. However, for those who
don't want to bother, here is an index which I have incorporated into a
new Club brochure to help prospective members assess 52-NOTES, and
identify back issues of interest. It covers V1N1 through V3N4, with
the V, N, and p designators omitted: Absolute Addressing 115, 173, 282,
2114, 333, 336; Advanced Programming Techniques 135, 154, 281, 2112,
343; Algorithms 145, 154, 164, 222, 254, 263, 293, 296, 2103, 2105, 323,
334, 342; AOS 134, 174, 232, 235, 2125, 312; Book/Periodical Reviews
141, 152, 171, 216, 224, 232, 236, 244, 256, 265, 274, 2124, 331, 345;
Built-in Functions 145, 163, 174, 221, 266, 273, 276, 282, 285, 296,
2116, 2121, 315, 321, 335, 345; Club News/Editorials 111, 126, 151, 164,
241, 246, 271, 274, 283, 2101, 2124, 2126, 313, 345, 346; CROMs 261, 275,
286, 2103, 2105, 2111, 2122, 2124, 313, 315, 335; Diagnostics 144, 173;
Display (non-fractured) 121, 136, 156, 222, 274, 2124, 314, 334; Dynamic
Code generation 123, 135, 344; Error States/Producers 112, 162, 174, 254,
266, 282, 2113, 2124, 313; Fibonacci Numbers 113, 222, 313, 333; Flags
123, 162, 233, 2102, 336; Fractured Digits/Display 113, 125, 146, 163,
166, 175, 222, 225, 234, 251, 2123, 311, 313, 316, 332, 346; Games 112,
116, 132, 143, 146, 243, 265, 2106, 325, 346; Hardware 122, 136, 144,
156, 161, 173, 174, 212, 216, 221, 225, 236, 253, 264, 2123, 331, 334,
336, 346; Indirect Addressing/Pointers 131, 154, 164, 215, 224, 276,
283, 315, 323; Input/Output 113, 115, 135, 215, 223, 236, 263, 281, 293,
295, 2102, 2103, 2106, 2126, 332, 334; Integer/Fraction Truncation 113,
124, 136; Interrupts 123, 176, 222, 314, 325, 336; Labels 115, 141, 173,
211, 254, 283, 286, 2114, 326, 332, 346; Mag Card Read/Write 124, 134,
144, 156, 194, 2105, 2122, 314, 322, 335; Matrix Calculations 134, 214,
223, 255, 283, 2125; Multiple Precision 255, 2101, 2121, 312, 313; Nota-
tion/Conventions/Definitions 126, 141, 161, 171, 226, 265, 292, 333;
Optimization 173, 2115, 2122, 222, 312, 321, 335; Partitioning 276, 283,
292, 2104, 2105, 2122; Printer 131, 136, 222, 284, 293, 295, 2102, 2105,
2121, 2126, 314, 321, 326, 346; Product Reviews 161, 171, 226, 253, 261,
331; Program Exchange (Club) 166, 246, 256, 275, 2116, 326; Pseudos/HIRs
124, 153, 173, 211, 245, 264, 274, 291, 2106, 2122, 315, 326, 336;
Random Numbers 231; Register Architecture 114, 121, 145, 163, 165, 172,
174, 213, 226, 254, 273, 2115, 335; Rounding/Precision 162, 171, 175,
221, 266, 2102, 341; Wipe-Out/Crash 122, 152, 163, 174, 175, 274.
52-NOTES V3N5p5
Periodical Review: TI-59 Newsletter, edited by Howard Gosman (754),
monthly since Nov 77, $24 per year.
Howard and a few associates have launched this newsletter as a
business venture (not a users club), claiming to target it to the non-
mathematician, "average" PPC user. The content and style of the first
5 issues characterize this periodical as oriented toward the non-prog-
rammer interested in business, finance, and gambling. Except for a
"beginners corner" which has addressed the use of built-in statistics
functions, there is little in the way of how-to programming tutorials
or other technical discussion. There is a fair amount of non-59 and
even non-TI PPC coverage over a broad range of generally non-programming
topics. Over 2000 subscribers are claimed, with growth to 5000 anti-
cipated.
Tips and Miscellany
56/57/58 Program Exchange: Dave Johnston (5) has a new catalog
dated 1 May 78 listing 32 math, 9 statistics, 10 operations research,
25 physics, 12 other physical sciences, 30 engineering and technology,
3 life and behavioral sciences, 13 games, 7 finance, and 1 general
info-test routine... programs written for the SR-56, of which 47 have
been translated for the 57, and 28 for the 58. 29 additional programs
were written for only the 57; 32 for the 58. Send Dave 20¢ and a SASE
for a copy of his catalog.
CROM Download into Used User Memory: A CROM download (Op 9) alters
only the steps which it fills, leaving other steps as they were before
the download. While the presence of user code remnants is unlikely to
affect execution of the downloaded CROM code, it can be a source of
confusion when the CROM code is listed. Fred Fitzgerald (252) sug-
gested (later confirmed by Tom Cox) that just such a situation caused
the Leisure Library-13 HIR confusion (V3N1p5, V3N2p6, and V3N3p6): The
HIR instruction at step 240 was a user-code remnant, not part of the
downloaded CROM. The safest way to avoid such confusion is to cycle
the power switch before downloading.
TI Service Manuals and Parts: David Swindell (877) reports that
detailed service manuals and some parts are available for the older TI
machines from Bootstrap Enterprises, Box 64, Richardson, TX 75080.
More on Memory Add-On (V3N3p4): Simon Hughes (837) reports that
he has successfully modified his 58, and has volunteered assistance to
Toronto-area members who need help. Simon's business phone is (416)
294-9838.
EE and Eng (57,58,59): Fred Fish (606) notes that INV Eng returns
either an EE or Eng display (hard or soft) to a turn-on display imme-
diately, while INV EE on a soft EE display doesn't remove the EE format
until another key is pressed. As the manual notes (V-8), the Eng
format continues to prevail following INV EE or CLR on an Eng display.
CROM Use of Flags for Print Suppression: Roy Chardon (515) noticed
that RE-2 uses flag 0 to suppress printing, but that no mention of this
is made in the RE manual. Further investigation reveals that RE-2 never
sets flg 0, but that RE-3 and RE-13 do, prior to calls to RE-2 to bypass
a Prt which would otherwise execute with printer connection. Similarly,
RE-4 uses flag 5 to suppress printing, but never sets it. RE-13 sets
flag 5 in the same sequence in which it sets flag 0 on the call to RE-2,
which in turn calls RE-4. Other members finding similar flag use in
CROM code are invited to share their discoveries, to help minimize the
unsuppressible printing problem, which is often a deterrent to CROM use
by user programs.
52-NOTES V3N5p6 (end)