Open Apple 1987 03 March 1987 Vol 3 No 2

Open-Apple fj Releasing the power to everyone. Reading AppleWorks data bases While a few of you are still hold outs, most folks in the Apple II kingdom have become very fond of AppleWorks in general and of the AppleWorks data base in particular. The great strength of the AppleWorks data base is speed. It sorts and retrieves data fast enough to make mainframes blush. Data entry and editing is smooth and quick. The amount of data that will fit in one record, of course, is limited to 30 different categories and what will fit on a single screen. The number of records that will fit in a single file depends only on what version of which companys RAMcard expansion software you have. For getting the data in your file out onto paper, AppleWorks allows you to define eight report formats. While these formats are sufficient for many types of reports, they pose significant limits for others. You can't, for example, dump your AppleWorks data onto a pre-printed form if more than 15 lines separate the highest and lowest areas you must fill out. If you use the AppleWorks data base to fill out continuous credit card forms (as we do around here), you have to include your company name and merchant number in every record; there's no other way to tell AppleWorks to print the same thing —even so much as a comma between city and state— on every form. There are a few fairly easy solutions to these problems. The mail merge features of AutoWorks and of the new AppleWorks 2.0 will solve both of the problems I've just mentioned. Another possibility is to open-apple-P(rint) your data base into a file, then read that file with your own program and manipulate the data any way you like. You could even update the data (for example, deduct today's sales from your toy store's inventory) and store the updated data in a DIF file. The DIF file could be loaded into AppleWorks as a new data base. The records in that new data base could be copied, using the clipboard, into the old data base and the old records deleted. This process is cumbersome, however, and not without some ill effects, such as the disappearance of hyphens from phone numbers and the inability to chronologically sort time and date categories. Another way to solve all these problems is to figure out how to have your own program directly read AppleWorks data base files. Those of you who are programmers probably realize that if you could get into the file itself you could manipulate and print out the data any way you wanted. You could even update the data and store a new AppleWorks-format data base file on disk. The possibilities are so immense that readers have been asking me to explain how to do it since I was struggling with volume 1. I've been reluctant to try, however. There's no straightforward way to read an AppleWorks data base file with Applesoft INPUT or GET commands. This leaves loading the file into memory and reading it with assembly language subroutines or with Applesoft PEEK loops. The problem with assembly language routines is that they take a long time to write and test and they are usually too long to publish. The problem with PEEK loops is that they are excruciatingly slow, For example, consider this little program: 10 HGR 2 FORC=1104 30 POKE 8192,C#S8 : FOR I = 8193 TO 16383 : POKE I,PEEK I-1 : NEXT 40 NEXT 50 TEXT Aboutall the program does is 32,768 POKEs and 32,767 PEEKs. Because it does them in the memory area devoted to the hi-res screen, with hi-res graphics tumed on, it also displays some really ugly patterns. Under standard Applesoft the program takes more than three minutes to execute ™ March 1987 Vol. 3, No. 2 ISSN 0885-4017 newstand price: $2.00 photocopy charge per page: $0.15 (190 seconds to be exact). On a Ilgs running in fast mode you get a boost factor of 2.7 —71 seconds total — still not very fast. While pondering these problems (after receiving the latest request for information on how to read an AppleWorks data base file), | remembered reading in the thin (but excellent) little manual that comes with the Beagle Compiler that compiled programs should use integer values whenever they can, because integers execute much faster than floating point values (the compiler considers an integer to be any whole number between -52767 and 32767, whether stored in an integer variable, such as 1%, or not). [wondered whether the compiler would speed up PEEK and POKE loops, which use such integer values. | compiled the above program and tested it Running on a normal Apple II, the program finished in 19 seconds — 10 times faster than normal. Ona ligs running in fast mode, the time was further cut to less than 8 seconds, which is 24 times faster than what we started with a paragraph ago and plenty fast enough to read a data base file. Sothis month I'm doing it. Here’s howto read an AppleWorks data base file. We'll save writing a new one for a later issue. File structure. The first thing you need to know is that AppleWorks data base files have three main parts. The first part is called the header. Among the goodies you can dig out of the header are the number of records in the file, the number of categories per record and their names, and the number of report formats that have been defined. There's also a bunch of other stuff there that’s important to AppleWorks but oflittle use to us, such as the format of the single-record and multiple-record screens and the current record selection rules. “WHY, WHEN T WAS YOUR, AGE I HAD TO TRAVEL Fit MILES OVER. A NON-COAXIAL CABLE To A LITE ard RING MACHINE IN A CORNER NEIGHBCRHCOD BAR? 3.10 Open-Apple The header itself can be split into two parts. The second part holds the names of the categories. The first part holds everything else. The first part is 357 bytes long. The second part has 22 bytes for each category thathas been defined. If all of the 30 possible categories have been assigned names, the second part is 22 * 30 or 660 bytes long, for a total length of 1017 bytes. The header can never be longer than this. (The only advantage to having data base files with less than 30 defined categories is that you save 22 bytes of file and desktop space per category not defined. The disadvantage, of course, is that should you ever need toadd a category to a file, your report formats will be deleted. It's far better, in the world of today’s expanded desktops, to always assign 30 categories to new files. Name the ones you have no current use for something like “+”, and use open-apple-L{ayout) to move them to a comer of the screen.) So much for the header. The second section of an AppleWorks data base file holds the report formats. Each report format uses 600 bytes. Thus, the length of this section can vary from nothing at all, if no formats have been defined, to 4,800 bytes if the maximum of eight formats has been defined. The third section of an AppleWorks data base file holds the actual data records, Records appear in the order in which you last sorted them. (Except that the first record always holds the file’s “standard values.’) Within each record, categories appear in the order in which they were defined with open- apple-N(ame), which is also the order in which the category names appear at the end of the header section. Reading a category. The record-category data is all scrunched together. This part of the file is similar to a sequential text file rather than being spread out in equal-length, mostly-empty segments as are random-access text files. The first two bytes of each record indicate how long the rest of the recordis. Next comes the data for the first category. The first byte of each category is a “control byte.” When this number is less than 128, it indicates how many bytes of data follow. Let's assume we have a variable called PNTR that points to the current control byte. The following subroutine will dig the data out of that control byte’s category and store it in a string variable called C$(N): 5100 REM read a single category’s data into CS(N) 5112 CBYTE = PEEK(PNTR) : PNTR=PNTR+1 : REM Get control byte. S162 CS="" : FOR I=PNTR TO PNTR+CBYTE-1 : CS = CS + CHRS(PEEK(I)) : NEXT 5190 CS(N)=CS : PNTREPNTR¢CBYTE —: REM Advance pointer to next category. S195 RETURN Line 5162 holds what | mean by a “PEEK loop.’ It begins by clearing a variable called C$, which will temporarily hold the category data. Then it loops the number of times required to dig the data out of memory. The PEEK(I) part of CHR$(PEEK(I)) tells us what value is at byte I, then the CHR$ part immediately converts that value into an ASCII character. You may find line 5162's “TO PNTR+CBYTE-1” puzzling. We have to subtract 1 from CBYTE because of the old “indexed from zero” paradox. If PNTRisat byte 100, and CBYTE says there are ten bytes of data, they would be stored in bytes 100 through 109. Looking in bytes 100 through 110 would retum eleven bytes of data, A clearer way to write the statement might be FOR I=1 TO CBYTE : C$=C$ + CHR$(PEEK(PNTR+I-1)). Writing it like that would slow down execution, however, because of the additional calculations inside the PEEK statement that would have to be done on each pass through the loop. Nowsuppose that the value in CBYTE is greater than 128. If CBYTEis 129, it means the next category is blank. If CBYTE is 130, it means the next tuo categories are blank. In other words, CBYTE-128 gives you the number of categories to skip. If CBYTE is 255, it means all remaining categories are blank and you have reached the end of the record. If there are no blank categories at the end of the record, there will still be a 255 marker, For example, a completely blank record takes up three bytes of space, no matter how many categories there are. The first two bytes indicate the number of bytes in the rest of the record (1—$01 $00 in hex) and the final byte is a 255, indicating all remaining categories are blank. A record with 30 categories, all blank but the last, would begin with two length bytes, then have a control byte of 157 (128 + 29), followed by a control byte indicating the length of the data in category 30, followed by that data, followed by a record- ending control byte holding 255. Let's add some lines to our previous subroutine to handle blank categories. “NB” is a variable that keeps track of the number of blank categories that should be skipped: 5100 REM read a single category's data into C5(N) 5110 IF NB > @ THEN S164 : REM In the middle of multiple blanks? 5112 CBYTE = PEEK(PNTR) : PNTREPNTR+1 : REM Get control byte. S114 IF CBYTE > 127 THEN S180 : REM Start multiple blank categories. Vol. 3, No. 2 5160 REM Category contains ASCII-string data. S162 CS="" : FOR I=PNTR TO PNTRSCBYTE-1 : CS 5164 GOTO 5190 = CS + CHRS(PEEK(I)) : NEXT 518@ REM Category Is blank. 5182 NB=C8YTE-128 5184 CS(N)="" : NB=NB-1 : GOTO S195 + REM CBYTE-S8@ is # of blank categories. 5190 CS(N)=C$ : PNTREPNIR+CBYTE S195 RETURN + REN Advance pointer to next category. One other detail we probably ought to consider comes up with categories that hold dates or times. AppleWorks uses a special storage format for dates and times to make them easier to sort. The first byte of a date category is 192 (SCO). The first byte of a time category is 212 ($D4). The first byte of any other kind of category is a low-value ASCII character, which will be less than 128, Date entries consist of six bytes. The first is the ID byte (192 or $C0). The next two hold the year in ASCII characters. The next holds the month, where an ASCII “A” means January, “B” means February, and so on up to “L” for December. The last two bytes hold an ASCII day-of-month, Time entries consist of four bytes. The first in the ID byte (212 or $D4). The next byte indicates the hour. An ASCII “A” means 00 (the hour after midnight), “B” means 01, and so on up to “X" or 23 (the hour before midnight). By adding the following lines to our previous subroutine, we can add the capability of reading dates. 1038 DIM HOS(12) + REN This array is for the nanes of the months. eb” : MOS (3)="Mar” NOS (6)="Jun” NOS (3)="Sep” NOS(11)="Nov” : NOS(12)="Dec 1044 MOS (7)="Jul” = 1046 M08(10)="Oct” = 5116 T=PEEK(PNTR) : IF 1128 THEN S16 : REM Date or time category? 5118 IF T=212 THEN 5130 5120 REM Category contains a date. 5122 YRS = CHRS(PEEK(PNTR+1)) + CHRS(PEEK(PNTR#2)) 5124 NOS = MOS(PEEK(PNIR+3)-64) 5126 OVS = CHRS(PEEK(PNTR+4)) + CHRS(PEEK(PNTR+S) S128 CS = NOS + ~~ + DYS +” ~ + YRS : GOTO 5190 5130 REM Category contains a tine. 5132 MS = “AM” : HR = (PEEK(PNTRé1) - 65) S134 IF HR > 11 THEN MS = “PM” : IF HR > 12 THEN HR = HR-12 S196 HRS = STRS(HR) : IF HR < 10 THEN HRS = “O" + HRS 5136 MIS = CHRS(PEEK(PNTR+2)) + CHRS(PEEK(PNTR+3)) S140 CS = HRS + "27 + MIS +” + MS: GOTO 5190 Line 5124 takes advantage of the fact that the ASCII codes for letters are sequential numbers. It tums the “A” that means January, for example, into a “1” by subtracting 64 from the ASCII code for “A” (65 or $41). “B” becomes a “2,and so on. The resulting value (1 to 12) pulls a month abbreviation out of the previously-defined array MO$(N). Likewise, in line 5132, the letter codes for the hours are turned into numbers between zero and 23 by subtracting 65 from the ASCII letter code. As written, these routines convert dates and times to strings that look exactly like what AppleWorks itself displays. With slight modifications you could arrange dates or times into any alternative format you might prefer. Reading arecord. Now thatwe have a routine that reads categories, itis a simple matter to write a routine that reads whole records. The following subroutine assumes only that PNTR points at the correct byte of the file and that the number of categories for the file has previously been placed in the variable NC. It also checks for the presence of the end of record marker and jumps to an error routine at line 5900 if it is missing. Since any error is quite likely a program error rather than a file error, the error routine prints some helpful debugging information (this particular program, of course, has been tested thoroughly and worked fine, of course, just before | sent it to the typesetter, of course—I include these lines, of course, in case your own program, of course, based on this one, of course, requires some fine-tuning, of course); 5000 REM read record’s categories into C$(1)...C8(N) 5010 RL = FN PK2(PNTR) : PNTR = PNTR+Z : REM RL is Record’s Length 5030 NB=@ 2 REN init # of blanks to 0 540 FOR N-@ TO NC-1 : GOSUB S100 : NEXT = REM get category data 5050 GOSUB S1@@ : IF CBYTE < > 255 THEN 5900: REM get SFF at end of record ‘5060 RETURN March 1987 5900 REM The File doesn’t look right--probably a program, not a file, bug. 5910 HOME : VTAB 10 5920 PRINT “I’ve encountered an error in the file’s structure” 5930 PRINT “in record “3R:” and category “;Ni”.” 5940 PRINT 5950 PRINT “The file buffer begins at “;88G;” and ends at “;BEN;”.” 5960 PRINT ” The buffer pointer is at byte “:PNIR:”.” 5990 END ‘The “FN PK2(PNTR)’ in line 5010 is a function that does a two-byte PEEK. The function is defined earlier in the program but later in this article. For more information on this trick see page 2.35 in our June 1986 issue. Memory buffers. Everything we've talked about so far assumes that somehow we've loaded at least a part of the AppleWorks data base file into memoryand that the variable PNTR points to the proper byte within the file. In order to do this we have to take care to set aside a block of memory we can use as a “buffer” and see to it that Applesoft doesn’t accidentally try to use the same memory area. We also want this memory area to lie in an address range less than 32768 ($8000) so that a compiled version of our program can PEEK with integers and run at maximum speed. One good place to put the buffer is between the Applesoft program and its variable tables. Figure 1 is a picture of how Applesoft uses your Apple's memory. Normally, Applesoft builds the variable tables adjacent to the end of the program image. By proper use of the LOMEM: command, however, we can move the variable tables and create an area of free space between the program image and the tables. This has to be done at the beginning of the program, however, before any variables have been used. (And in order to work with the Beagle Compiler, which doesn’t update the PRGEND— program end—pointer at bytes 175-176 quite right, it has to be done without referring to PRGEND.) How about: 1000 REM Program initialization 1010 LOMEM: 16384 + PEEK(105) + PEEK(106)*256 : REM Create 16384-byte buffer. 1030 DEF FN PK2(ADR) = PEEK(ADR) + PEEK(ADR+1)%256 : REM 2-byte peek function. 1032 DIM CNS(30) : REM This array is for category nanes. 1034 DIM C$(30) : REM This array is for the information in categories. EN BBG points to the beginning of our buffer. EM BEN points to the end of our buffer. EM PNIR paints to our position in the buffer. : REM BYTE points to our position in the file. ‘$0000 1054 PNTR = BBG 1056 BYTE=0 $2000 $3000 ‘$4000 ‘$5000 ‘$2000 Sinple variable table $7000 stant of arrays prray variable 107-108, $88-$8¢ se000 table end of variables ‘STREND ‘$9000 Fee 100-110, $00-$ee «strings event $090 ual ly overwrite this area. End of stri os $8000 bith 11-112, $66-$70 [String storage area 1 \ $0000, Start of strings tEnSi2 <HIRER: > 115-116, $73-$74 Basie Istart of pos _— ‘tio std pointers; 7 Applesoft Program = usu! !y $9600, i tor ft con vary Rent e Lad Area sloatg.> ‘err Figure 1 Open-Apple 3.11 The PEEKs in line 1010 look up the current location of the variable tables; LOMEM: and the addition of 16384 move them $4000 bytes away from the end of the program. Lines 1030-1036 set up the two-byte PEEK function and dimension some arrays we'll need later. Lines 1050 to 1056 set up some pointers to the beginning and end of our buffer and to our position in the buffer and in the file on disk. In line 1052 we actually divide the $4000 bytes we have set aside into two buffers—only one will actually be used in this program, but we'll need the other in a couple of months when we give this Program the ability to write AppleWorks files, too. One of the beauties of ProDOS is the ease with which any type of file can be loaded into memory, even in small pieces. The upcoming subroutine for loading sections of AppleWorks data base files uses the BLOAD command, the A(ddress) and Lfength) parameters that DOS 3.3 programmers are familiar with, and the T(ype) and B(yte) parameters that can only be used with ProDOS. The T(ype) parameter allows any type of file to be BLOADed. BLOAD is not restricted to binary files, as with DOS 3.3. Here, the file type we want to load is “ADB.” The B(yte) parameter allows you to begin loading a file into memory at some position other than the beginning of the file. Setting B to 1000, for example, will cause the BLOAD to skip the first 1000 bytes of the file. This capability is absolutely necessary —without it you can’t access files that are larger than the buffer. With it, on the other hand, you can load the file into the buffer in pieces. Here's our subroutine: 5500 REM load section of file into buffer 5510 BYTE=BYTE+(PNTR-88G) : PNTR=BBG 5520 PRINT CHRS(4);"BLOAD”;FS;”, TAO, L8192, 9°;88G;", B°;BYTE 5530 RETURN Line 5510 calculates a value for the B(yte) parameter by determining the distance between our pointer, PNTR, and the beginning of the buffer. Then it adds that difference to the previous B(yte) value. When we initially load the first section of the file, these variables will cancel each other out and equal Zero, SO we'll begin BLOADing at byte zero. Later BLOADs will essentially move the byte that PNTR is pointing at from the end of the buffer to the beginning of the buffer. Somewhere we need to check to see if PNTRis nearing the end ofthe buffer, Agood place to do this is right after line 5010, where we find out the length of the record we are about to read. We need only add the record length to the PNTR and see if the result is beyond the end of the buffer. If so, we should reload the file. This line will take care of all that: 5020 IF PNTRIRL => BEN THEN GOSUB S500 : REM Does record extend beyond buffer? The header. After we have the beginning of the AppleWorks data base file BLOADed into our memory buffer and before we start reading the actual data embedded in the records, it's necessary to dig a few important pieces of information out of the header. The header length is stored in the first two bytes of the file. This length does not include the two length bytes themselves, however, so we need to add two to the result to arrive at the full header length. The number of records in the file is a two-byte number and can be found at bytes 36 and 37 (where the first byte of the file is byte 0). The number of categories in the file is stored at byte 35. The number of report formats is stored at byte 38. And the category names are stored in 22-byte segments beginning at byte 357. Each name begins with a length byte and is followed by up to 21 ASCII characters. Here's the instructions for digging all this information out of the header: 1100 REM Load first section of file and dig stuff out of the header. 1110 GOSUB $500 : REN load File 1120 HL = FN PK2(PNTR}+2 : IF HL > 1017 THEN 5900 : REM header length 1122 NR = FN PK2(PNTR+36) + REN i of records in file 1124 NC = PEEK(PNTR+35) THEN 5900 : REN # of categories 1126 NF = PEEK(PNTR#38) THEN 5900 : REM of report formats : IF NC > 30 : IF NF >8 1140 PNTR=PNTR+357 1142 FOR N= @ to NC-1 1144 C$="" : FOR I=1 TO PEEK(PNTR) : CS = CS + CHRS(PEEK(PNTR+I)) : NEXT 1146 CNS(N) = CS : PNTR = PNIREZZ 1148 NEXT If you modify or amplify this program and you are pressed for space, you can make the buffer smaller. Don’t make it smaller than 1K (1024 bytes), however, or you might not be able to read the whole header in one chunk, It's also possible, though unlikely, that a single record could hold slightly more 1 REM get category names 3.12 Open-Apple than 1K of data, so a 125K buffer is probably a safer minimum size. To make the buffer smaller, change "16384" in lines 1010, 1050,and 1052 and"8192" in line 5520. After you've finished reading the header, PNTR will point to the first byte after the header. The next section of the file holds report formats, which we want to skip over completely. The following lines will do that: 1150 PNR = PNTR + NFX6Q0 : REM Skip over 600-byte-each report formats. 1152 IF PNTR => BEN THEN GOSUB 5500 : REM Do formats extend beyond buffer? That pretty much takes care of reading AppleWorks data base files. You now have everything you need to read such a file from an Applesoft program. Here are some additional program lines, however, that dress-up this demo a little so that you can easily read any AppleWorks data base file without actually running AppleWorks. CAUTION: THIS PORTION OF THE PROGRAM DOESNT INCLUDE THE PROGRAM LINES ENBEDDED IN THE ACCOMPANYING ARTICLE, WHICH ALSO MUST BE TYPED IN TO MAKE THE PROGRAM RUN. THE ORDER IN WHICH YOU TYPE THE LINES NAKES NO DIFFERENCE, JUST DON’T SKIP ANY. DON’T ENTER LINES 10 THROUGH SO. 100 REM *#* Open-Apple’s ADB Reader *** 101 REM by Tom Weishaar, Feb 1967 1020 CLEAR : REM Restart point for reading another file. 1036 DIM 18(30,2) : REM This array is for category TAB positions on screen. 1050 PRINT CHRS(4);"PRI3” : PRINT : REM 80-colunn screen required. 1062 VTAB 10 1064 PRINT “lthat is the name of the Appleviorks database file you want to see?” 1066 PRINT Vol. 3, No. 2 1130 FOR N = @ TO NC-1 + REM Get screen positions. 1132 TB(N,@) = PEEK(PNTR+186+N) : REM screen sequence left-top to right-bot 1134 TB(N,1) = PEEK(PNTR+114+N) : REM horizontal screen position 1136 TB(N,2) = PEEK(PNTR+1S0+N) : REM vertical screen position 1138 NEXT 1160 REM Just for fun, drau an Appleliorks-Like screen for display. 1162 PRINT “File: “;RIGHTS(FS,20) : PRINT : PRINT “Selection: All records” 1164 VTAB 1 : POKE 1403,28 : PRINT “Open-Apple’s ADB-READER” 1166 VTAB 7 : FOR Isl 10 8 : PRINT ~ 1168 VTAB 23 : FOR Il 10 8 : PRINT ~ 1170 PRINT “Press spacebar to see next record. 1200 REM Get records and display then on screen. 1210 FOR R=0 TO NR 1220 GOSUB $000 : REM this loads CS(N) with record’s data 1230 REM The rest of this Just displays the data on the screen. 1232 VIRB 6 : POKE 1403,0 1234 IF Re@ THEN PRINT “Standard Values for this file:” : GOTO 1236 1236 PRINT “Record “;R;” of “:NR: CHRS(29) : REM chr8(23) clears line 1238 FOR N= @ TO NC-1 1240 VIAB TB(N,2)+1 : POKE 1403,TB(N,1)-1 1242 PRINT CNS(TB(N,O)-1);“: “3C8(TB(N,O)-1); CHRS(23) 1250 NEXT 1260 VTRG 24 : POKE 1403,37 : GET AS : PRINT AS; : REM wait for key 1270 NEXT 1300 REM End game. 1310 HONE : VTAB 12 1320 PRINT “That’s all the records in “;FS 1330 PRINT 1340 PRINT “Would you Like to see another? <y/N>“; : GET AS : 1350 IF AS="Y" OR AS=“y* THEN 1020 1360 END PRINT AS : HOME Miscellanea ProDOS 8, version 13, released to developers in mid-January and mentioned here last month, mistakenly includes a BRA (branch always) machine language instruction in a critical piece of code. This is a significant problem, because BRA isn’t supported by the original 6502 microprocessor. Consequently, this version of ProDOS causes bad things to happen when run on Apple Il-Pluses and unenhanced Iles. If you BLOAD PRODOS, TSYS, A$2000, then the bad instruction appears at $4CCD. A BEQ (branch on equal) would work just as well here, so you can fix things with a POKE 19661,240 and a BSAVE PRODOS, TSYS, A$2000. Another significant problem I've encountered while running the newer versions of ProDOS on earlier machines has to do with interrupts. ProDOS 111 and earlier versions disabled interrupts. An alien static- or bad-luck- caused interrupt signal caused no problem with these versions of ProDOS unless some software that used interrupts had enabled them. Newer versions of ProDOS, on the other hand, leave interrupts enabled at all times in order to support some newfeatures of the IIgs. If you use these versions on an earlier computer, an alien interrupt will lock it up with the message “INSERT SYSTEM DISK AND RESTART” at the bottom of the screen. At least that’s my diagnosis of why I've seen that maddening message more in the last month than in all my previous incarnations. I've reverted to using ProDOS 111, patched as described in our November issue, on everything around here except the Iigs. I suggest you do so as well until further notice. The problems that Apple’s SCSI card had with the IIgs have been solved, There is a revision B EPROM now available for the card (part # 341- 0112-B); contact your dealer for details. You might be surprised at the kind of things you can get on continuous forms. The advantage of continuous forms, of course, is that they are easy to process through any printer that has a tractor feed, which is just about any printer nowadays. Besides the continuous credit card forms mentioned in this month's main article, you can also get such things as mailing labels, envelopes, invoices, purchase orders, statements, letterhead stationery, checks, and even Rolodex cards on continuous forms. Spend a few minutes wandering around your local office supply store to see what they have that you could use. Wouldn't life be easier if you designed a “labels style” report format for AppleWorks that could print your address-phone number data base on Rolodex cards? Ifyou live in the U.S., read this and act quickly. U.S. federal tax forms for 1986 are available on AppleWorks spreadsheet templates for $23.95 from Personal Financial Services, P.O. Box 1401, Melville, NY 11747 516-261-8652; for $42.45 from Sky Computer Resources, P.O. Box 204, Portland OR 97207 503-234-7291; or for $50 from Island Computer Services, 3501 E Yacht Dr, Long Beach, NC 28461 800-826-7146. All three packages include the main 1040 form, as well as the forms for Employee Expenses, Depreciation, Itemized Deductions, Interest and Dividend Income, Profit from a Business or Profession, Capital Gains, Supplemental Income, Self-Employment Tax, and the Married Couple Deduction. All three packages produce IRS acceptable print-outs for all forms except those that are color-coded (the IRS does want you to use green-bar or lined paper, however). Personals package, called 1040Works, also includes forms for Farm Income, Income Averaging, Credit for the Elderly, Sale of Residence, Child Care Expenses, Moving Expenses, 10-Year Income Averaging, and Alternative Minimum Tax. If you have enough memory for a 256K AppleWorks desktop, pay $3 more and ask for 1040Works-; you'll get all this stuff in one large spreadsheet. Sky's disk also includes form 1040A and the schedule for Farm Income. Sky can provide any other form or schedule printed by the IRS for an additional $5 each. Island's disk also includes the forms for Child Care, Moving Expenses, Income Averaging, and Credit for the Elderly, as well as tax planning templates for 1987. Island is the only company of the three that accepts credit cards. If you don't live in the U.S., (or Canada, Mexico, Australia, or New Zealand) read this. Because of a bad scale at our mailer's, our December issue, mailed near the end of November, went out with only enough airmail postage for 1/2 ounce. Since the newsletter actually weighed slightly more than that, the U.S. Postal Service seems to have kindly forwarded that issue to you by surface mail. Except for that issue, we have paid for 1 ounce of airmail on each of your newsletters each month. If you have received any other issues by surface mail, or ifyou receive an issue by surface mail in the future, please send the empty envelope back to Sally Dwyer at our Overland Park address so she can figure out where our intentions have gone astray. Ask (or tell) Uncle DOS Man talks, Apple listens Iwas sick of reading how wonderful your newsletter is in all the computer magazines. I was sick of getting little notices in all my new Beagle Brothers software telling me you were still alive. 1 was sick of getting postcards in the mail asking me for a good reason not to subscribe to Open- Apple. So I took you up on your offer of a free issue. And now, damn it, I'm a member of the insidious “Gee, one suggestion alone was worth the entire cost of the subscription” club. Ok, enough of that. Let me tell you what I'm doing with my Apple Ile. At the very least, I'd like to lean what others might be doing in a similar vein. I'm the statistician for a group of baseball fanatics in what we call the Duckball League. (About the name, well, its a long story,) Anyway, we draft major league players onto our teams and compete with each other in a number of statistical categories. And since I'm the one with the computer, I'm the statistician. Funny how things work out that way. Tuse AppleWorks and about a 74K spreadsheet to keep track of everything. Normally, it takes me about two hours a week to enter the data on the almost 500 players in the league. The most difficult part is looking back and forth from the tiny type in USA TODAY (our bible), to the keyboard, to the screen, and back again. | figured I could speed things up by eliminating one of those three elements. Enter voice recognition. After checking the few manufacturers of Apple voice recognition systems | could find, I decided on Intravoice II from the Voice Connection in Irvine, California. It consists ofa main plug-in circuit board, a microphone, and a couple of smaller boards into which you plug several of the Apple Ile’s ICs. All in all it's pretty easy to install. The voice input module itself is quite easy to use and can be trained to recognize just about any word. That word can then be used to as a substitute for any keystroke or string, It supports open- and solid-apple commands as well, so its ideal for my use in entering Duckball stats into the AppleWorks spreadsheet. I have encountered one and only one problem in my entire use of Intravoice Il to date. When using a phrase (“PinPoint”) to substitute for a solid-apple-P keystroke to invoke the PinPoint Desktop Accessories, AppleWorks crashes into the monitor. I'm talking dead. Once I'm in PinPoint, voiced solid-apple com- mands work fine, But somehow when AppleWorks is active it doesn’t work. Rex Creekmui Grand Rapids, Mich. Your letter leaves me speechless. The address of The Voice Connection is 17835 Sky Park Circle, Suite C, Irvine, CA 92714, (714) 261-2366. The elusiveness of relative power In Compute!’s Apple Applications Vol. 4 (Fall/ Winter 1986), William Mensch, the one-man 65816 progenitor, states that the 65816 (and presumably the 6502) runs in such a manner that one of its bus ycles is equal to four of anybody else’s (page 18). He states that a 65816 running at 6 megahertz is “equivalent to the IBM PC running at 24 megahertz” and “a 2- to 5- megahertz Apple has the same kind of performance as an 8 megahertz Macintosh." This information, if correct, is of critical importance toa typical hobbyist, such as myself, ready for his next generation computer. The marketing power of this alleged fact is exponential. Many times in the last three years I have pondered to myself something like, “My Ile is a wonderful machine but I wish I had the speed and power of the 8088”. It seems to me that the 6502's longevity and the allegiance it commands may be due as much to its unsung power and speed as to marketing forces over the past ten years. Maybe it's not ancient and obsolete but instead was ahead of ts time when created and is still incredibly capable in its present form. If this conjecture has some credibility, then it follows that even though the 65816 is panned by blue loyalists as a 16-bit introduction at the dawn of a 32-bit era, the brute-force power and sophisticated characteristics may well approximate, parallel, or even outshine the favored sons of Mac and Charlie. Regardless, the technology for building a San Francisco skyscraper is quite different than that for building the Golden Gate. Bridge. Both are indispensable. Steve Cranney Fallon, N.Y. Microprocessors are complex devices. Just as you can control the superiority of quarter-horses over Thoroughbreds by shortening the race, microcom- puter loyalists can devise benchmarks that show their favorite microprocessor is the “most powerful.” That's why the benchmarks we published here (December 1986, page 2.88) were the standard benchmarks Byte has been using for years—we didn’t want to be accused of favoritism. Yet that's not to say we would have been so eager to publish the results if the 6502 and its progeny hadn’t come out looking so good. Dennis tells me Mensch's comments (great interview, by the way, congratulations to Compute!) are based on the fact that the 65. series uses a scheme called “pipelining” that allows it to grab the next piece of data it will need while it’s still working on the last piece and on the fact that all instruction codes in the 65n0xx series are just one byte long. This places some limits on the number of different instructions but allows faster execution. The general philosophy of the 65x series is that the processor spends less time finding out what it is supposed to do and more time actually doing it. This translates into faster execution of common tasks at the expense of the inclusion of more powerful commands, such as multiply or divide instructions. This same philosophy has been used to develop a new breed of “reduced instruction set” computers that actually execute only a few instructions but do so very quickly. The IBM PC-RT is an example of this kind of machine. Open-Apple 3.13 Ina practical example, the 65816 is faster than a 68000 (at the same clock rate) for simple load accumulator and save accumulator operations, which would favor the 65816 in a benchmark based on that ability. The 68000 should kill the 65816 in a math benchmark, however, because the 68000 has math instructions that the simpler 65816 instruction set does not. Nonetheless, if a benchmark doesn't use the 68000's math operations and instead does math “manually,” as the 65816 does, then the comparison would favor the 65816. I suspect this is essentially what happened with the Byte benchmarks we reported. Dennis recommends you take a look at the series of books Adam Osborne has written on microcompu- ters for further information. Inthe final analysis, the most important thing toa hobbyist shouldn't be a microprocessor's power, anyhow. Youshould be looking fora computer witha good, reliable, overall design, with good software that does the kinds of things you want to do, and with good documentation (very important, but often overlooked). An example of the importance of docu- mentation is the first commercially available 16-bit microcomputer, the Texas Instruments 99/4. TI wanted to develop all the software for the machine itself and locked hackers out. No documentation, no hacking; no hacking, no programs; no programs, no customers. What was the advantage of owning the most “powerful” microcomputer of its day if you couldn't figure out how to do anything interesting with it? RGB and TV too I have a Sony KV-1311CR monitor hooked to my Apple ligs. Its a 13-inch, cable-ready, remote-control TV, with analog and digital RGB inputs and video inputs and outputs. It has a dandy picture, both in normal TV use, and when used with the GS analog output. KV-1131s can be had mail order for less than the Apple monitor (the latest price I've seen is $399). | also like it because I can hook up the GS to the analog input, a II-Plus to the video input, and cable TV to that input, and have them all there for use in one neatllittle package. The pinouts in the Sony book are as if you are looking at the mating cable connector, not at the jack on the TV. On the monitor, pin 1 is at the bottom, towards the back. On the ligs, pin 1 is on the top, towards the on-off switch. Incidentally, the numbers don’t match the pin numbers molded on DB-15s, or at least not on the one | used. Here’s how to run the wires: Sony 6S Functicn Pin Pin Red Ground Red Signal Green Signal Green Ground Blue Signal Blue Ground Composite Sync You can also enable the IIgs audio output, but this is less straightforward. The Sony requires +2.5 to 5 volts on pin 34 to enable the audio input. Otherwise, you'll get TV audio all the time. Apple neglected to put +5 volts on its jack. However, there is +12 volts. You can have a technician build you a voltage divider out of a couple of resistors to cut back the +12 volts, which comes out of pin 8 on the IIgs. The audio signal 3.14 Open-Apple comes out of pin 11 on the Ilgs and goes into pin 24 on the Sony. An alternate option is to use the voltage divider to automatically switch on the video RGB when you turn on your computer. To do this, hook the +5 volts to pin 35 on the Sony. Chris Arndt So, you can geta “free TV” by carefully selecting a IIgs monitor. The March 1987 issue of Consumer Reports compares 175 color television sets from 21 different companies; nine of these sets reportedly have RGB inputs. The nine are J.C. Penney model 2220, Magnavox models RF4254WA and RG4378BK, Quasar models TT6290XE and TT6298YW, Sanyo models AVM210 and 12C700, Sears model 42701, and Sony model KV-20XBR. There are, no doubt, others — the set you have, for example, isn’t among the 175 sets listed. Slashed zeros and 8 bits Ihave spent many hours trying to get my Imagewriter II to print slashed zeros, I've tried using the control codes given in the manual; it works fine except that I get a double-spaced printout. As | read the printer manual, the printer codes required for slashed zeros are: Slash on: £SC D Control-@ Control-A Slash off: ESC Z Control-@ Control-A What am | doing wrong? Richard E. Breininger FPO Miami, Fla. ESC D and ESC Z allow you to change the Image- writer dipswitch settings under software control. ESC Dis used to turn switches on. ESC Z is used to tum switches off. There are two eight-switch groups of switches, called A and B, for a total of 16 switches. After the ESC D or ESCZ the Imagewriter expects to see 16 bits of data that tell it which switches to tum on or turn off. Apple’s sequence for turning on the dipswitch for slashed zeros is (hex values) “1B 44.00 01”. This causes switch 1 on dipswitch B to be tumed on. This tells the Imagewriter to use slashed zeros and slashed zeros we have. Unfortunately, however, Applesoft takes the liberty of setting the high bit of each byte printed (for a complete discussion of this problem see “Control- D(efeated)” in our December 1986 issue, pages 2.84 86 and “A bit too many” in April 1986, page 2.24). Thus, the actual values sent are “9B C4 80 81” Consequently, we actually turn on two additional switches—switch B-8, which is not used by the printer (no problem) and switch A-8, which adds a linefeed after carriage return if itis on (Presto! Instant double-spacing). One answer is not to send the command sequence through Applesoft —see our December issue for the details. In this case, however, we don't have to get so elaborate. We can just turn the linefeed option back off after turning slashed zeros on. In other words, after sending the codes ESC D control control-A, also send ESC Z control control. Your printer will see “9B C4 80 81 9B DA 80 80" and will begin giving you slashed zeros without extra linefeeds. Here's a good tip for those of you having problems similar to this one—it comes from a letter in our May 1985 issue, page 39. Tum your Imagewriter off, pressand hold down on the linefeed button, and turn it back on. It will now print the hexadecimal code for each character it receives instead of doing normal printing. This feature can be quite useful for diagnosing printer command-code problems. PRINT TAB alternatives Tell me how to get a PRINT TAB statement to look like it does on your Apple IIc screen when you're printing to an Imagewriter II. Help—I'm a desperate woman! Robin Boscia Pittsburg, Penn. I have a problem with an Imagewriter II, Super Serial Card, and enhanced Ile: my Applesoft program has tabs and the printer will not go to the correct columns. Tabs range from column 1 to column 103. Can you help? David E. Brewer Cincinnati, Ohio Amazing. I've spent all morning playing with PRINT TAB on an enhanced Ile with a Super Serial Card and it hardly does anything right. After trying all kinds of combinations, including printing with the screen display in 40 and 80 columns and printing with video echo on and off (control- |), I finally found a combination that worked: before turning on your printer, PRINT CHR$(21) to the screen to turn the Apple 80-column firmware off. Then, after tuning the printer on, PRINT CHRS(9);"T E” (space required between T and E) to enable the Super Serial Card's “Basic Tab” command, which I don’t remember ever hearing of before this moming’s search through instruction manuals, That, anyhow, made this program work (using ProDOS): 10 HOME : PRINT CHRS(21) : REM Turn off 8-colunns 20 PRINT CHRS(4);“PRUL” £* : REM Turn on ‘Basic Tabs’ RINT SPC(3) 31; : NEXT : PRINT 40 FOR I=1 TO @ : PRINT "1234567890"; : NEXT : PRINT 50 FOR I=1 10 4 60 PRINT TAB(30); “ROW”; TAB( 40) ;“OKR”;TAB(S0) ; “MRO” 70 NEXT 80 PRINT CHRS(4); “PRES” 82 ENO A significant problem with this trick is that the Apple llc serial port command set doesn't include the “control-I TE” command. Using PRINT TAB while both the 80-column firmware and the printer are tuned on doesn’t work at all. This is because something, probably the firmware but I'mnot sure what, forces the values in CH (byte 36 or $24) and OURCH (byte 1403 or $57B) to zero. CH holds the horizontal cursor position for 40-column firmware and OURCH the position under 80-columns. But Applesoft expects these locations to also reflect the horizontal position of the printer. Yet when both the printer and 80-column mode are on, both bytes always hold zeros (that's why an initial PRINT TAB works fine but later ones work just like PRINT SPC). Any readers who can provide more information on what the bug is or how to make TAB work right are encouraged to write. Meanwhile, here are two ways to do tabbing without using PRINT TAB, Let's say you want to print a table in three columns, Assume the widths of the columns are stored in the variable array CW(c). The ‘Stuff you want to print in the columns is stored in the variable array T$(r,c). In this case, try the following: 10 FOR R=1 TO NR: REM For rou 1 through # of rows 20 FOR C=1 TO NC : REM For col 1 through # of cols 30 CS*LEFTS(TS(R,C),CM(C)) : REM truncate 40 PRINT CS; SPC(CH(C)-LEN(CS)); 50 NEXT : PRINT 60 NEXT Vol. 3, No. 2 Inline 40 we print the data, then use the PRINT SPC function to fill out the column with spaces. We determine how many spaces to print by subtracting the length of the string just printed from the column's width. To use this trick we have to make sure the data isnever wider than the column. Ifitis, the subtraction will produce a negative number and the program will stop with an ILLEGAL QUANTITY ERROR in the ‘SPC function. That's why line 30 truncates everything to the width of the column. The biggest disadvantage of this method is that all these string manipulations will slow your program down—you may find your printer is waiting on the computer rather than the other way around. The second alternative is to use your printer's tabs. This presents two problems. First, you have to figure out how to set the tab stops. Then you have to figure out how to get the control-I character, which most printers use for a TAB command, through your interface card and out to the printer. For the Imageuriter, the command sequence “ESC (“tells the printer you are sending a list of tab stops, This list comes immediately after the command and is a sequence of three-digit numbers, separated by commasand terminated with a period. The numbers will be sent as ASCII characters, so we don’t have the high bit problem mentioned in the last letter. To set tabs at columns 30, 40 and 50, for example, use PRINT CHRS(27);"(030,040,050." (If you're having trouble with the conversions of things such as ESC into things such as CHR$(27), just memorize the “ASCII Control Code Rosetta Stone” on page 85 of the November 1985 Open-Apple.) The Imageuriter also allows you to clear a list of tab stops—just use “ESC )” instead of “ESC (". You can clear all tab stops with “ESC 0”. When setting tabs, the left-most print position is tab stop 1. The maximum permitted setting depends on the width of the character set you are using. Once you have set a tab stop, however, it remains in the same absolute position on the page, (at least with the Imagewriter) even if you change character widths or the position of the left margin. There are two ways to sneak a control-! through an interface card. You have to sneak them through because most cards use control: as a “wake-up, here comes an interface card command” code. Consequently the cards eat any control-Is they see rather than passing them on to the printer. Some cards, including the Super Serial Card, will pass a single control-I on to your printer if you print two of them in succession. Otherwise, you have to change the card's command code to something other than control-I. Do this by sending a control-I followed by any other control-character. If you do this, however, change the command code back to control-I before you turn the printer off. Try this program to see how an Imagewriter handles tabs with your interface card: 10 PRINT CHRS(4) ; “PRI” 15 PRINT CHRS(2: 20 PRINT CHRS(27);”(030,040,050. 30 FOR I=1 10 8 : PRINT SPC(3);1 40 FOR I=1 10 8 : PRINT “1234567890: : REM clear all tab stops NEXT : PRINT XT: PRINT 50 TAS=CHRS(S) : Z5=CHRS(26) S2 PRINT 1882S : REM change cnd code to ctrl-Z 54 GOSUB 90 5G PRINT CHRS(27);°L020";"Left margin at col 20.” 58 GOSUB 90 60 PRINT CHRS(27);"L000";"Left margin at col 0. 62 PRINT CHRS(27);"P";"Proportional type” 64 GOSUB 9@ March 1987 66 PRINT CHRS(27);“Q";"17 chars per inch.” 68 GOSUB 90 70 PRINT 26;T8S : PRINT “TAB with tuo ctrl-Is.” 22 TOS=TaS+ TBS 74 GOSUB 90 80 PRINT CHRS(4);“PRHI~ 82 END 90 FOR I=1 104 92 PRINT T8$;“Nice”;1BS; “coluens,”;78$; “right?” 94 NEXT = PRINT : RETURN GET out of text files | have been working on a program that creates downloadable characters for my Imagewriter printer. Thad it save the characters in a text file. When | tried to read the data back | discovered something unusual, Myread routine was similar to this: 10 GET AS 20 A = ASC (AS) 3@ GOTO 10 Ifyou type this in and runt, itworks fine most of the time. The problem arises when you try to read a $00 (or $80 as it is stored in the file). Ifyou try control-e in the above program you get an ILLEGAL QUANTITY ERROR in line 20. Any idea why? Can | make it quit doing that? Steve Carder Liberty, Mo. GET will not read an ASCII $00 or $80 (control-6) Character; it interprets it as a null string. The GET does retum to your program, however, so here's how to detect an ASCII $00: 10 GET As 15 IF LEN(AS) = @ THEN A = @ : GOTO 30 20 A = ASC (AS) 30 GOTO 10 Nonetheless, a text file is probably not the best choice for saving data like this —text files always set the high bit. Try POKEing your characters into a buffer and BSAVEing them. Or, if you insist on a text file, convert the characters to ASCII numbers while saving them with PRINT STRS(A) and read them back with INPUT AS : A=VAL(AS). This will probably quadru- ple the size of your file, however. Multiplan to DIF In regard to the question in your November 1986 issue (page 2.79) about converting Multiplan files to ‘SuperCalcfiles, a recent issue of Computist includes a program by D.W. Walkey that converts Multiplan's SYLK files to DIF files (issue #37, page 20). I've used the program to move several spreadsheets to APPENDS: J.D. Holdeman N. Ridgeville, Ohio Computist's address ts P.O. Box 110846, Tacoma, WA 98411 DIF files, of course, will move only the values from one spreadsheet to another, not the underlying formulas. Editing data files Inyour November 1986 issue (page 2.77), Lawrence Pratt mentions that he edits programs and data files, even random access files, with Apple Writer. Can you give more information about how this is done? Is this abetter method than using GPLE or simply a different way of doing the same thing? Thomas E. Militello Rancho Palos Verdes, Calif. Everyone has his or her favorite method of writing programs. GPLE and similar programs are “line- oriented” editors— you edit one line at a time. Many of us who are addicted to “screen-oriented” word processors, however, keep what we know about line- oriented editors and ox-drawn vehicles on the same index card. To use aword processor to write programs, simply make sure you enter a retum at the end of each line. Save the program in a text file (from AppleWorks, open-apple-P(rint) the program to “A text (ASCII) file on disk.”) Exit your word processor, enter Applesoft, and EXEC the text file you have written. Uncle DOS will type your program in for you. For more on this, including how to get programs you have already written into text files you can edit, see the May 1985 Open-Apple, page 36. If you happen to use Apple Writer for program editing, CondiCom’s program OpenAppleUriter is very handy (December, page 2.87). Another possibility is Program Writer from the Software Touch (Sep- tember, page 2.62c), which isa full-powered, screen- oriented program editor. Using a word processor to edit data files is similar to editing program files. In order to edit random- access files, however, the file should be completely filled with blanks or other characters before being used the first time. For much more information on this subject, see my column in the May 1984 Softalk, Page 164. Only exit FOR-NEXT at NEXT The insertion sorts printed in your November 1986 issue (page 2.80) had a common erroneous method of exiting a FOR—NEXT loop. Itshould read: first routine: 940 IF A(E) <= T THEN E=E1 : GOTO 960 second routine: 950 IF A(K,D%(E)) <= A(K,T) THEN E=E1 : GOTO 965 Before you exit a FOR—NEXT loop, you need to close out Applesoft's handling of that loop to free up the stack. The amended lines will set the loops’ index variables to their maximum values and GOTO the loops’ NEXT statements, so that Applesoft will close out the stack handling for the FOR—NEXT before continuing, Craig Willford Whittier, Calif. You are correct. | talked about this problem in the April 1984 Softalk, page 52. Using GOTO from inside a FOR—NEXT loop to a line outside the loop leaves Applesoft expecting a NEXT that never comes. Exit like this from eleven loops with different index variables in a row and you'll get an OUT OF MEMORY error. Applesoft won't say so, but it means it’s out of stack memory. Asa practical matter, many programmers use the same index variable for all loops, so they rarely see the bug. Applesoft fixes the stack automatically when a loop, or any other loop that uses the same index variable (E in this case), is re-executed. None- theless, there is a limited amount of stack memory in the Apple II (256 bytes), consequently programmers should try to keep it free of muck such as unresolved FOR—NEXT loops. More TransWarp experiences | read with interest the letter from Pat Mamet in your January issue, page 2.96, with reference to Open-Apple 3.15 problemswith a TransWarp board. | used an Ace 1000 with a TransWarp and Peachtree's Back to Basics accounting package with no problems. Recently, 1 upgraded(?) to the Ace 2100. Upgrading my Back to Basics program to the Ile configuration and using the TransWarp, | lose all data on the second drive. Removing the TransWarp, everythings perfect. Thank the Lord | copy data regularly! I've called AE before about the TransWarp and came away feeling like a stupid ass from the condescending attitude. | don’t think the ProDOS patch would work because Back to Basics is a DOS 3.3 program, so my TransWarp is sitting on the shelf. Frank Drew Seminole, Fla. I have been reading your letters about ProDOS zapping disks. Although I never experienced that, I have experienced three zaps of some significance to me and I wonder whether you have heard of similar complaints. Briefly, I have zapped three disks when I tried to disable my TransWarp, using the escape key, after power-up. The three disks were all copy-protected and non-ProDOS — the reading program Smart Eyes, a Moebius scenario disk, and a program called MicroTest from Harper and Row that | use in my sociology class. My computer has PinPoint's Ile upgrade kit and a Checkmate 768K memory board. Larry Davis Bedford, Texas We've received a couple of other letters reporting problems with TransWarps since publishing Mamett’s letter. My own Ile even started acting flakey last week (spreadsheet cells in columns to the right of my work area having nonsense formulas in them, characters I hadn't entered appearing in unusual places on the screen, computer locking up) and all the problems went away when I replaced my Trans- Warp with an older SpeedDemon. The big problem, of course, is that once you're used to working with a TransWarp, it is very difficult to go back to something that's slower. Another difficulty is that the problems come and go—Dennis put my TransWarp in his Ile and hasn't had a bit of trouble. I suspect the root problem is that the TransWarp pushes the Ile hardware to its limits. My guess is that weak chips or noise in the computer, rather than in the TransWarp, are causing most of these problems. Another potential culprit is dirty contacts on the TransWarp’s edge connector. These are just guesses, however. RAM, compiler comparisons I recently called Applied Engineering and ordered the latest version of their desktop expander software so I could compare it with the software that came with my Checkmate MultiRAM CX card (for the Apple IIc). As far as | can tell there does not seem to be any problem running Applied Engineering's $10 software with Checkmate’s card, except for the AppleWorks Desktop Expander itself, which doesn't work. Other comparisons between the two disks have left me favoring Checkmate’s $5 software. Checkmate’s RAMdisk, called /MRAM, is configured for slot 3 drive 1 and leaves /RAM in slot 3 drive 2 intact. AE’s RAMdisk, on the other hand, expands the the size of /RAM in slot 3 drive 2. | prefer Checkmate's way of creating the RAMdisk with a SYS file rather than a file that must be run from an Applesoft program as with AE's software. 316 Oper-Apple AB'sdisk did havea nice program that uses auxiliary memory for aa one-pass disk copy (5.25 disks only). | don’t see any advantage to using it, though — since it must first format the disk it is going to use it isn’t any faster than Copy II Plus, Both disks have very similar auto-load programs for moving files to the RAMdisk. | did not compare the two except to notice that both seem very easy to customize. | always use MouseDesk's auto-load feature so | don’t need another auto-load program. I have two compiler bugs to report. Interestingly enough, they were both discovered while using two different compilers on the same program and both problems were similar enough to affect the same program line. The first one is in the Beagle Compiler (version 10) and is illustrated in the following example: 10 D = 23: DS = STRS(D): OX = VAL(DS): PRINT OX The above statement will print 23 with Applesoft and print 0 when compiled. The other bug is in Micol BASIC (version 2.2) and can be demonstrated as follows: 10 0s = Dx = VAL(DS): PRINT D& This statement will print 9 with Applesoft and print 1024 with Micol BASIC. I've sent letters to each company and | would expect they will implement fixes shortly. The Micol Open-Apple a ‘s writen, edited, published, and © Copyright 1987 by Tom Weishaar Business Consultant —_ Richard Barger Technical Consultant Dennis Doms Circulation Manager _ Sally Tally Business Manager Sally Dwyer Most rights reserved. All programs published in Open-Apple are public domain and may be copied and distributed without charge. Apple user groups and significant others may reprint articles from time to time by specific writen request. Requests and other editorial material, including letters to Uncle DOS, shouldbe sent o: Open-Apple P.O. Box 7651 Overland Park, Kansas 66207 U.S.A. Published monthly since January 1985, World-wide prices (in U.S. alias; arma delvery included ato adaional charge) $2610" year, $44 for 2 years; $60 for 3 years. All single back issues are Currently available for $2 each, bound, indexed editions of Volume 1 and Volume 2 are $14 95 each. Volumes end with the January issue; an index for the prior volume is included with the February issue. Please send all subscription-related correspondence to Open-Apple P.O. Box 6331 Syracuse, N.Y. 13217 U.S.A. Subscribers in Australia and New Zealand should send subscription correspondence to Open-Apple, c/o Cybernetic Regarch i £76 Maven Raa raran, C31 AUSTRALIA 1en-Appie is available on disk trom Speech Enterprises, P.O. 021586 Pbuston Texas 7270 713-461 668) Unlike most commercial software, Open-Apple is sold in an ‘unprotected format for your convenience. You are encouraged to ‘makeback-up archival copies or easy-to-read enlarged copies for your own use without charge. You may also copy Open-Apple for distribution to others. The distribution fee is 15 cents per page per copy distributed WARRANTY AND LIMITATION OF LIABILITY. | warrant that most of the information in Open-Apple is useful and correct, although drivel and mistakes. are included from time to time, usually unintentionally. Unsatisfied subscribers may return issues within 180 days of delivery for afullrefund, Please include anote from your parents or children confirming that all archival copies have been destroyed. The unfulfilled portion of any paid subscription will be refunded on request. MY LIABILITY FOR ERRORS ANO OMISSIONS IS LIMITED TO THIS PUBLICATION'S PURCHASE PRICE. In no cease shall | or my contnbutors be liable for any incidental or consequential damages, nor for any damages in excess of the fees paid by a subscriber ISSN 0885-8017 Source Mal: TCF238 Printed in the U.S.A. CompuServe: 70120,202 BASIC bug can be worked around as follows: 1008 = "9": = VAL(DS): DX = Anyone who uses the PinPoint Desktop Accessories may want to remove some accessories from the list or rearrange them. This can be done by loading the text file PINPOINTPROFILE from the install disk; then arrange the file the way you want it; then save the file back to the disk and run the install program. You may even be able to install your own favorite binary programs (perhaps a game) as accessories by adding itto the list along with its length and highest memory address. David Stevens Eden Prairie, Minn. 1: PRINT DX Our experience and a growing pile of letters from our subscribers indicate Checkmate also has much better technical support. The Beagle Compiler is up to version 2.0. In addition to a fix for the STRS bug, the latest version lets you automatically store strings and arrays outside of main memory. This gives you more room for programs and much more room for data. The extra memory supported includes the second 64K bank of a 128K Ile, a lic, or a Ilgs; auxiliary slot RAM cards from Applied Engineering and Checkmate; and slot 1-7 memory cards from Apple, Applied Engineer- ing, and Cirtech. Memory, Apple Writer gotchas In your discussion of the various kinds of RAM in the December 1986 issue, you said “Interestingly, none of these switches (HIRES/LORES, STORE4O/ STORE80, PAGE1/PAGE2) actually do anything to the current display that appears on your monitor.” Well, that's almost true. If text or graphics page 1 is already being displayed, then tuming on STORE80 will not change the display. But if page 2 is being displayed, turning on STORE8O will change the display to page 1 as the PAGE1/PAGE2 switch takes on its new meaning, On the older unenhanced Apple Ile, this little interaction can cause a minor display annoyance in certain programs — specifically programs that display text on hi-res graphics page 2 and rely on monitor routines to help with the text bookkeeping. This is because a few monitor routines in the unenhanced Ile, including the scroll routine, tum on STORE80 for a brief moment even with 40-column operations. While STORE80is on, the display changes to graphics Page 1, causing a brief “flicker” on the screen. The enhanced Ile and IIc do not access STORE80 during 40-column screen operations and so do not have this problem. You suggest that the “best” use of the auxiliary bank 64K of memory or an auxiliary slot memory board is as a RAMdisk. Do you mean from an applica- tion programmer's point of view or a user's? Asa user, [find both the built-in ProDOS RAMdisk and the third- party auxiliary-slot memory board RAMdisks less than convenient to use. Sometimes the most conve- nient way, or in some cases the only way, to get from one applications program to another is to press control/open-apple/reset to reboot. But the contents of these auxiliary memory RAMdisks are lostwhenever you reboot. You have to be careful to use any Quit options provided when moving from program to program. In addition, for the auxiliary-slot memory boards, you have to runa special program every time you start up in order to “install” the RAMdisk driver code into memory. I've talked to too many customers who believe that they should be able to cold-boot any Vol. 3, No. 2 ProDOS program and have that program automatically recognize their auxiliary slot memory board as a RAMdisk. Since their memory board works so well with AppleWorks, they believe there's a bug in any program that can’t do that. You mention in the article that Apple Writer doesn’t follow the documented protocol for disconnecting the auxiliary memory RAMdisk. Here's another inter- esting and obscure little tidbit—Apple Writer also diddles in silly ways with the Super Serial Card. The printing set-up file that comes with Apple Writer includes the setting “CR1”, which means Apple Writer will supply its own linefeed character after each carriage return. But the printer cards on most Apples are already set to supply their own linefeed after carriage return. They need to be set that way for printing from Applesoft to work. So how does Apple Writer avoid double spacing? It first checks if the printer interface is either a Super Serial Card or a IIc printer port. If it is, Apple Writer pokes new values directly into the screenholes reserved for the printer interface to turn off the interface card's linefeed. Now Apple Writer can add its own linefeeds, and every SSC or IIc owner is happy. Well, almost...Apple Writer doesn't repoke the old values back into the screenholes and the Super Serial Card doesn’t forget previous pokes unless reset is pressed. The other day I printed a document using Apple Writer, then used “control-Q J” to exit “gracefully.” I switched to another ProDOS program (never needing to press reset), then tried printing something out to the printer from this program. The printer did not advance the paper. Line after line was printed one on top of another, since the Super Serial Card would not. issue linefeed characters after carriage retums. The change that Apple Writer had made was still in effect. Remembering these problems, | pressed control- reset to force the Super Serial Card to reinitialize itself, then printed again. This time it worked fine. Phil Thompson Portland, Ore. So that’s why my spelling checker prints differently every time I use it. Apple is always harping at developers to put everything back just like they found it (no problem here, | agree with them), but Apple's own software is as bad as any at not following the rules, The ProDOS quit feature, combined with program selection spftware, has created a trend toward not rebooting machines between applications. Thus, itis becoming |more and more important that each application program undo any changes it makes toa computer gr its attached devices. My comments about it being “best” to use large memory cards as RAMdisks were aimed at both Programmers and users. It's true, however, that a significant|problem with the auxiliary-slot cards is that the software that makes them work must be “installed|every time the user reboots. However, neither Applied Engineering's nor Checkmate’s aux- will lose their contents during a reboot as long as bank Os locked out. I'm still looking for a universal, licensable RAM disk driver that would recognize any type of RAM and convert it all into a large RAMGdisk if no other driver was already installed. Commercial developers could include such a driver within their software to make memory management totally transparent to users. The advantage to devel- pers, of course, is that all RAM cards (to say nothing of hard disks and whatever storage devices the future holds) look and act alike, even though they are in reality quite different.