8bitfiles.net/archives

home *** CD-ROM | disk | FTP | other *** search

/ 8bitfiles.net/archives / archives.tar / archives / genie-commodore-file-library / Information / CAT5-940127A.SFX / 01cat5 < prev next >

Wrap

Text File | 1990-02-12 | 33.5 KB | 815 lines

CBM-ED [e.g.bell] (Forwarded) I can add a couple of suggestions to yours JBEE. In your first line, with the dsave, I always precede that with a scratch command and follow it with a dverify command. But after the REM, I put a quotation mark followed by enough deletes to pull all but the command back to the start of the line and delete the REM and the line number. I usually include an 'x=08' or something as the first statement on the line, and a ,u(x) after the 'scratch', 'dsave', and 'dverify', so that I can easily change device numbers in all 3 commands with 1 or 2 keystrokes. It usually works out to about 8 deletes after the quotes. *note* this will *NOT* work if your filenames have upper case letters in them. I do something similar w/my routines, but I renumber them just before I save them to disk, then incorporate the first number into the file name. For example, if I have a routine to display a directory, I might renumber it starting with line 1000 in increments of 1 line number. then when I save the file, I'll call it 1000 directory. I have a COMBINE command built into my C128 chips, and use that to build up a foundation of routines for new programs. Just renumber when I'm finished. Each routine on the 'routine disk' starts with a different group of line numbers, which is why I can combine them w/o fear of overwriting anything, and since I renumber in increments of 1, I have a *lot* of space to maneuver. BTW, that inclusion of deletes can work for any line you use to clean things up with after a test run. I use one to restore the character set pointer and one also with the FIND command, also in my C128 chipset. Your suggestion about the routines reminded me of the disk I have dedicated to that purpose from TotalRecall that speeded up development of the support programs. It is a source of much of what I have entered into that Snippits topic in Cat 21. ------------ Category 5, Topic 3 Message 23 Fri Jul 24, 1992 C128.JBEE [* Sysop *] (Forwarded) The only thing I don't like about scratching old files is sometimes I need them :D Also, I keep a time and date log of all the files created both for development, legal, and accounting purposes. So it literally doesn't pay for me to scratch them :) Of course if you are using a 1541/71 and saving 140+ block files every ten minutes scratching the old ones will probably save a ton of disk space. Even at $0.25 a piece you could go through some dollars fast :) The combine command sounds pretty interesting, did you burn your own routine into a ROM? ------------ Category 5, Topic 3 Message 24 Fri Jul 24, 1992 HOWIE-CBM (Forwarded) JBEE, The Basic program listing, and structure is terrific! You have several features that I will now incorporate into my template for use with new projects. They all seem so easy to implement. I don't know why I never thought of them. Very helpful info! :) Howie ------------ Category 5, Topic 3 Message 25 Fri Jul 24, 1992 HOWIE-CBM (Forwarded) Ed, Yes! Tell us more about your custom chips!! And especially COMBINE. :) For short combines, I've used that ole: LIST, LOAD, and enter <CR>'s. For big combines, and where something gets inserted into the middle of something else, BTP's buffer. :) Howie ------------ Category 5, Topic 3 Message 26 Fri Jul 24, 1992 CBM-ED [e.g.bell] (Forwarded) John: get ready to blush. The COMBINE command in my C128 is indeed from a set of custom chips from Free Spirit, a ROM replacement that I was sold on when I read the review many years ago in TC128. Course, you were not the publisher then. The chips also have a FIND, CHANGE, and MERGE. The MERGE gets a little flakey sometimes, not sure why. I have a Super Chip (ha!) from I forget who that is really a piece of garbage. Anyway, the COMBINE function just appends programs to the end of whatever program is in memory. Then I just renumber the file. It's great. as a matter of fact, I use FIND and CHANGE so much, I don't think I'll get JIFFY DOS any time soon cuz I would lose that in their chip upgrade. The chipset also fixes some bugs like the shift-q, and I believe the relative coordinates in the graphics commands. There are other commands I have long since forgotten because I don't use them. I think there is also a numeric base conversion function but since I have that in the BASIC language, the MLM, and probably my Mach 128, I"m not sure about that. I believe there is also a nice set of FKEY definitions,but there is some contention among chips in my C128 so I"m not sure which chipset has the priority in that regard. Know the ones I like are not the ones I have, but it is a nice chip set. At $50, it was not cheap though, and I don't know if it is still available, but for the FIND and CHANGE alone, it was worth every penny! ------------ Category 5, Topic 3 Message 27 Sat Jul 25, 1992 C128.JBEE [* Sysop *] (Forwarded) Yea, I remember that review ;) ------------ Category 5, Topic 3 Message 28 Sat Jul 25, 1992 M.MCCORMIC10 [Mike Mc-] (Forwarded) I have uploaded the genealogy prigram that we are discussing and started a topic for it in Cat.3,Topic #27. The files are #'s 9900,9901,9902,9903,9904,9905. All are SDA. Mike Mc- ------------ Category 5, Topic 3 Message 29 Sun Jul 26, 1992 CBM-ED [e.g.bell] (Forwarded) I grabbed them yesterday Mike. They dissolved ok, but file 3, I believe, has a couple of duplicate files at least. I started over on a separate disk after 2 failures for that reason. The filenames as uploaded don't reflect the fact that they are SDA's either, for general info. I didn't have time to run the prgs, and will now have to move them around on disks first. >>sigh<< but this is a *massive* program in terms of support file count, etc. Can you give me some kind of idea what files I *need* to run the prog and duplicate the problem. I'd rather not have to deal with all of those files if I don't have to. ------------ Category 5, Topic 3 Message 30 Sat Oct 31, 1992 M.MCCORMIC10 [Mike Mc-] at 07:29 EST Ed- I didn't realize that you had also answered my question in this topic.You can eliminate; overlay,notewriter,and files that have those names. I will repeat the problem.This program will do all that I ask "EXCEPT"create a record or chart for the irst great grandfather person # 8. That is where I get "string to long"error.All the great grandparents work fine. Could there be a missing module? This is everything that I have. Mike Mc- ------------ Category 5, Topic 3 Message 31 Sat Oct 31, 1992 CBM-ED [e.g.bell] at 08:53 EST I just posted a long answer to your other post, Mike. See my post and then we can go from there. I have to dig up my copies of all those files. When dissolving them, I found at least 1 or 2 duplicate files in different archives and was never sure if they were different, or what. Also, this puts another thing into the equation. Your data. Could you drop a copy of what you are entering here or in my mailbox (CBM-ED) so I can enter things exactly as you are entering them. I may or may not be able to duplicate your results if I am not using the same data. But let me see what I can do... and give me every detail of what you have done. I've never used a genealogy program so I am starting from square one on every front. ------------ Category 5, Topic 3 Message 33 Tue Jan 19, 1993 R.KNOP1 [Rob Knop] at 00:43 EST I echo Bruce's sentiment that the Hacking Magazine is a good place to see source code. Some of the articles there are pretty technical, and useful even for folks who are old hands at ML. Recently, the editor, Craig Taylor, started up a series where he's writing a "Space Invasion" game, which looks like it could be very useful for someone who knows the basics of ML but is unsure how to do anything with it. Another good place for source code with explanations are the old Transactor magazines! -Rob ------------ Category 5, Topic 3 Message 34 Sun Jun 06, 1993 GULLIBLE [Jon] at 16:53 EDT Can't think of a better place to put this... I'm working on a sequel to my PI DEMO. A demo which has the first 18,000 digits of PI in the scroll. I now have a file listing the first 1.25 Million digits of PI and I would like to come up with a compression method that will let me put as many digits possible in available memory...probably about 40K. So far, the best I can get is a 2.4:1 compress ratio. It works like this: Take the sequence: 123456789012 123 = 0001111011 456 = 0111001000 789 = 1100010101 012 = 0000001100 These 40 bits equal 5 bytes: 00011110 11011100 10001100 01010100 00001100 Thus, 12 digits in 5 bytes or 2.4:1 Can anyone think of a better compression method? The reason I ask is that 0-9 is only 10 different characters and there are 256 in a byte. 25:1??? I know this is a weird request... :) But, I'd appreciate any help or suggestions. Thanks. ------------ Category 5, Topic 3 Message 35 Sun Jun 13, 1993 C128.JBEE at 09:01 EDT Gullible, I have been thinking this over and believe I have found a solution, short of compression, that might work, especially with a series of numbers such as PI. Which is special in the way the numbers are (not;) sequenced. It takes 3 bits to represent a digit 0-9, but how about representing the difference from one digit to another instead of using a trinary solution? Example: Start the data flow with the first number then start reading everything in according to bits If the first bit is zero it means to use the next bit to add or subtract 1 from the number in the A,X, or Y register. Then display that number. 0000 0000 -> first bit=0 so use the next bit to determine if you should add or subtract, it is zero so subtract -1 from the A reg 0000 0010 -> first bit=0 so use the next bit to determine if you should add or subtract, it is one so add +1 to the A reg If you have a zero following the nine you would add one but write a zero to the register. So you would have to sacrifice speed for compression by doing comparisons to a table of rules that says if A is greater than nine (CMP) then A = 0 If you have a sequence of numbers such as 4345 that would require 12 bits at least but using this method only 8 bits! Now what happens if the first bit encountered is a 1? Then use the next three bits to branch to a table that keeps spliting the table of remaining numbers in half. 01x3x56789 0135 6789 Bit #2 true false 0135 6789 / / Bit #3 =1 Bit #3 =0 true false true false 01 35 67 89 / / / / Bit #4 =1 Bit #4 =0 Bit #4=1 Bit#4 true false true false true false true flase 0 1 3 5 6 7 8 9 The only draw back is the size of the table vs compression gained and the logic needed to keep track of where you are in the current byte (bit wise). I am not even sure it would compress the file better than using your method. But this is all I could think of taking into account the way PI is represented and not knowing the odds of a digit having or not having a number lower or higher than it, following it. Though I am sure the occurance of such is probably high. Hey, that in itself would make an interesting project :) ------------ Category 5, Topic 3 Message 36 Sun Jun 13, 1993 GULLIBLE [Jon] at 15:58 EDT Thanks for the response. I hadn't thought about using a difference method. I am a bit confused as to how 4345 can be represented as 8 bits. If you have two bits that say subtract 1, then how do you know if those two bits mean subtract one or are the first two bits of a 3-4 bit sequence? I will look over your post in greater detail. I do believe that using math will yield the best results. ------------ Category 5, Topic 3 Message 37 Sun Jun 13, 1993 C128.JBEE at 23:17 EDT If the first bit is unset (0), you know you are going to add or subtract with the bit following it. If the following bit is set you add one, if it is unset (0) you subtract one. So a sequence of numbers such as 4345 (remember we started with a number three) only requires two bits per digit because we are representing the change, not the digit. Since the sequence 4 3 4 5 up one ____ ___ up one down one ______ ______ up one looks like that it would be represented in binary as 10100010 ----(4 up one) (3 down one) (4 up one) (5 up one) 8 bits! You use the first bit to determine if you are going to be using two bits or four bits to determine the next number. So you will always have at least a 2x compression over the storing of just the numbers is ASCII form. ------------ Category 5, Topic 3 Message 38 Wed Aug 25, 1993 T.RUTTER2 [Troy Rutter] at 19:17 EDT Hi again people, well, its time to jump on the bandwagon and TRY (crossing fingers) to finish my SIDplayer. As many of you have followed through the Ultrasid Player topic under Sight & Sound, my SID player plays the individual parts of the sid archives, msw, pil, wpl, sal, slr, etc... without having to dissolve them. The only thing holding me back is the words display. Right now the conversion time from in memory to display is awfully sloooooooow. Heres the scoop: For some odd reason, the creators of Singalong songs (Kermit and Joe I think?) decided to store the word files in standard ASCII instead of PETSCII. So... heres what happens in my program as it stands now. (Simplified of course) Im doing this as a basic source around ML subroutine... so bear with me... get location of words file, assign this to variable: WD create array wd$ get character from memory compare - does it need to be converted? If so, convert If not, store as is in WD$(x) if peek(wd)=carraige return then x=x+1:go start over if peek(wd)=0 then start music, start wds display. This takes forever. I have also tried two other ways: 1) The "Index table" as it is referred to by ML experts. This works... but still bothers me... slow, plus i have no idea what im doing. 2) Display without storing in wd$. This is faster, but it is not guaranteed to display fast enough to keep in time with the mus flags. If anybody can take a look at some SID player source code and help me out, i would appreciate being enlightened on how to display the words. It seems like most sidplayers do some kind of banking so its automatic. I dunno. I have the source and object code in PAL (i think) of sidpic 3.4 if that would help. I REALLY want to get this sucker done before the next SID-Fest. Thanks, Troy ------------ Category 5, Topic 3 Message 39 Wed Aug 25, 1993 CBM-ED [e.g.bell] at 21:26 EDT Troy: TR> I have also tried two other ways: 1) The "Index table" as it TR> is referred to by ML experts. This works... but still bothers TR> me... slow, plus i have no idea what im doing. 2) Display TR> without storing in wd$. If you are insistent on doing this display in BASIC, the 'Index Table' is the fastest possible if done correctly. What you are doing is creating a lookup table for conversion from ASCII to PETSCII. You use the numeric value of the ASCII character as an index into the table, which eliminates all IF statements, the REAL time hogs in a BASIC program, from the equation entirely. I would recommend doing the whole operation in ML, which would assure the needed speed, and if you shelled to it from BASIC, you would still be in control of the timing of the display. An added plus is that it would take far less memory to do in ML than it would in BASIC, particularly when you consider the elimination of the arrays, another BASIC time hog. TR> If anybody can take a look at some SID player source code and TR> help me out, i would appreciate being enlightened on how to TR> display the words. It seems like most sidplayers do some kind TR> of banking so its automatic. I dunno. I have the source and TR> object code in PAL (i think) of sidpic 3.4 I will be happy to do that for the opportunity to get hold of the sid player source code. What you are requesting seems to me to be an extremely easy thing to do. However, when is the SID-Fest. I will have to get your source code quickly to get it done for you because I am leaving for vacation tomorrow night. On the upside, I can't imagine what you want taking any more than about 5 minutes to code once I have the source code. If you can't work with this time-frame, I hope you will still share the SID Player source code, which I've been looking for for a long time. I'm not planning anything that would 'compete' with your work, and not for the C64 at all. But how does all of this sound? I already have the lookup-tables in files from other programs.... You get the code uploaded tonight I may be able to have your solution first thing in the morning when I get home from work. (The above assumes we are talkingML source code, tho your mention of PAL format pretty much indicates that) ------------ Category 5, Topic 3 Message 40 Wed Aug 25, 1993 R.KNOP1 [Rob Knop] at 22:25 EDT 5-3 Troy- what exactly is done in the "compare" step where you decide if the string needs to be converted? Is this something that detects whether the string is PETSCII or ASCII, or is it done on a character-by-character basis? One thing that will help is to write a short ML subroutine to do your conversion for you... this will tend to make things faster. Once in ML, my favorite way to do these things it the "Index Table" method. Basically, you need to build a 256 byte table. Let AV be the ASCII value (known) of the character, and PV be the PETSCII value (desired) of the character. Then, the AV'th byte in the table is PV, assuming a properly built table. In other words: ldx #AV ; or wherever you find the ASCII value, put that in .X lda table,x At this point, .A will hold the PETSCII value, and you can stick it wherever you want it. -Rob ------------ Category 5, Topic 3 Message 41 Wed Aug 25, 1993 T.RUTTER2 [Troy Rutter] at 22:37 EDT Ed: Sounds great, except for one small detail. When Kermit made his source code available for SIDpic, he left the following in the file description: "This code requires the use of Craug Chamberlain Sidplayer M/L (version one) that was especially modified to work with my player. If you want to work on this and add more to my player, I cannot help you beyond providing this code abd pointing you to the M/L attached to SIDPIC V3.5 - I cannot distribute Craig's M/L as it might offend him. If you want it though, you CAN figure out from my Source Code where his M/L loads into, and take it out of SIDPIC V3.5." So, the actual ML that PLAYS the music is still in executable form only. I would still appreciate it, however, if you took a look at the source code. Especially routines such as: frommem convert and etc. PS.. the SIDfest is not until 1994. -Troy ------------ Category 5, Topic 3 Message 42 Wed Aug 25, 1993 C128.JBEE at 22:48 EDT Since I have the exclusive rights to the book, I do not mind the passing of the code back and forth if it means getting something accomplished that will increase the sales of the book. ------------ Category 5, Topic 3 Message 43 Thu Aug 26, 1993 CBM-ED [e.g.bell] at 10:00 EDT Does that mean, John, that you do have the source code for the SID player? ;) Troy: I looked at the source code you sent me today, and I can honestly say I can't understand why Kermit made such an extensive use of the .bas pseudo-op in PAL! It is confusing reading, but I did find something you should find interesting. The routines you want to convert and print the word strings seem to me to already exist in the code. The pertinent routine is FROMMEM in line 352. It first calls the routine called ROUTINE, which moves the variable pointers for I presume A1$ into the variables at $8c-$8e. These are used as pointers and general information into the string. Note the definition of A1$ in line 11. This seems to me to be defining the string in memory for use by the FROMMEM routine. It receives the string data when you call FROMMEM from the BASIC routine. Note particularly lines 161-175 (esp. lines 166-175). If you study this routine, you should be able to either use it or see how it is used. It seems to me to do exactly what you want done tho, and in ml to boot. The routine CONVERT at line 410 in your source file is a conversion routine. I did not study it closely, but based on what you say, I assume it converts A1$ data to the proper format. It makes this change to the string directly. Without running this, the code seems to me to: set the word address in lne 168, poking the low and high bytes into 113 and 114 ($71/$72 used in FROMMEM) I'm not sure what line 166 is doing, but I bet it is turning something off, probably an interrupt (I just checked... that is exactly what it is doing. See line 833) 167 is also assigning LA the value held in .x(low) and .y(high). Line 169 calls FROMME, which loads the next string into A1$, and then puts the length of that string into A. Line 170 puts the string into t$, necessary since you NEVER want to mess with A1$ or you would goof it up as far as length!!!! It assigns it the value of left$(a1$,a) which pares it down to its actual length. Line 171 apparently sets up the pointer for the next string an also stores that in WA. Line 172 sets up the FROMMEM routine again, and line 173 calls it, doing as described above. Line 174 converts a1$ for printing and stores it into w2$ and updates the address pointer. At this point, I don't know what the difference is between t$ and w$, but you could find that out by doing run-time displays (after reassembling the program with the appropriate instructions) to see what is in them with each iteration of the routine. Bottom line is that the routines you need are already written and available for your use. Give them a try and see if I'm not right. Now, do you have access to Kermit Woodall? I contacted him YEARS ago about using Wxmodem in BellTerm,tho I never did. I don't have his address even if I did have a presence of mind with him. Based on what John said about the book rights, it seems you have a basis for asking for Craig's player source code if you are willing and have that kind of access to him, and I would consider it a personal favor if you did! egb ------------ Category 5, Topic 3 Message 44 Sat Aug 28, 1993 T.RUTTER2 [Troy Rutter] at 13:46 EDT Ed, all, unfortunately, I have tried to use those subroutines in my programs and it locks up... i think it is bad memory management... here is what I have: Main basic program: 7169-10026 ML (frommem,routine,convert) 12288-12378 SID files load in at 15000 sid routine: 49152-53152 And it locks up after i use modlink to put them all in one file. Im wondering if modlink iteself uses some of that memory in there. Also, i wonder why the basic program loads at 7169... isnt it sposed to load at 2048 or something? BTW.. modlink = Modlink by Robert Stoerelle (MALAKAI) Any idears? ------------ Category 5, Topic 3 Message 45 Sat Aug 28, 1993 C.DRUTIS [Monsoon] at 23:15 EDT Greetings. This seems as good a place to ask as any... Could someone tell me if it's possible to run an ML subroutine within Bank 1? I want to put a text editor within Bank 1, but no matter what I do with the MMU register or the PCR's, the program refuses to work properly (or at all) within Bank 1. It works fine when I load it in Bank 0, though. For Bank 1, I begin with the following code: LDA #$4E ; corresponds to %01001110 which SHOULD be : Bank 1 RAM STA $D500 ; Kernal and IO . . . STA $FF00 Of course when I use it in Bank 0 I use LDA #$0E. Any help would be apprecitated. [Apptest 0000321.A] ------------ Category 5, Topic 3 Message 46 Sun Aug 29, 1993 J.JONES36 [LOADSTAR] at 00:09 EDT YOu could try STARLINKER, on LOADSTAR #105. It can link up to 140 blocks worth of BASIC and ML. Just remember that with SL, your program MUST have a CLR before any variables are declared. The trouble you're facing though really isn't because of MODLINK. I see this as I'm typing. MODLINK and STARLINKER are meant for BASIC prgs that run from $0801. Your BASIC PRG is running from around $1800. Here's what you do: Turn off your computer for a few secs. This clears memory in case you're going to crunch the resulting program. Turn it on and write a one liner at $0801 10 POKE44,HI:POKE43,LO:POKE HI*256,0:RUN Where HI is the HI byte of your program's location and LO is the Low byte. DON'T RUN IT YET! Now type: LOAD "YourProgram",8,1 Your program MUST have been saved from where it runs, not only for proper LOAD address, but for line links. If you have JiffyDOS, DON'T use the %filename substitution since this won't set SOV pointers. Now type: SAVE"combo",8 This will SAVE the one liner you just typed in, a big gap of nothingness, then your program at $1800. When RUN, this program will switch control to the program at $1800 and RUN it. THIS file will link properly with MODLINK or STAR LINKER. You can include any ML or graphics data between these areas, but LOAD the highest data last. If you have font data at $800, use a monitor to load the fon, then the one line program, then your BASIC program, then save them all in one file. Your BASIC program will have to restore the 8-16 bytes of corrupted font data when it's first RUN. Or you can just have some ML copy an entire font from elsewhere. I link at least two programs per month. Usually they run at $1000. I have no prob. ------------ Category 5, Topic 3 Message 47 Sun Aug 29, 1993 C128.JBEE at 00:43 EDT It sounds like a Basic 7.0 shell (C-128) using the original SID ML (C-64). ------------ Category 5, Topic 3 Message 48 Sun Aug 29, 1993 R.KNOP1 [Rob Knop] at 00:43 EDT Troy- I don't think you want to link these various sections into one file. It looks like there are gaps between where the various sectios want to go. I don't know how this linker works, but if it just links the files together into one program, the resulting program will load sequentially in memory starting at the load address, not putting the various parts where they ought to go. If modlinker takes care of this, somebody please correct me.... Re: the starting address, the usual start for BASIC programs is indeed 2049. It sounds like the code was moved up for a relocated start-of-BASIC (e.g. to make room below the start of BASIC for a custom character set, a trick I used to use way back when). How do you want this program to function after it has been linked? -Rob ------------ Category 5, Topic 3 Message 49 Sun Aug 29, 1993 R.KNOP1 [Rob Knop] at 00:44 EDT Monsoon- I assume that you are issuing a "BANK 1" command before loading the ML program and SYS'ing to the address? If not, then there's the simple answer :D What are you calling in the way of BASIC and KERNAL subroutines? I am not sure, but I believe it possible that some of the Kernal subroutines may only return to bank 15, or at any rate a configuration with RAM 0 rather than RAM 1 visible. Hopefully, and probably, this is not the case, and the Kernal routines are friendly, but it is possible. It is more likely with some of the BASIC routines. It's been years since I've directlly called any BASIC or KERNAL routines from RAM 1, and the only reason I did that was because I was doing it from within GEOS, and in GEOS it is usually not necessary to directly call BASIC or the Kernal. Any idea just where the code is dying? What is the manner in which it dies? Ideally, ML things should work in RAM 1 just as they do in RAM 0. Curious, why are you locating the program in RAM 1 as opposed to RAM 0? -Rob ------------ Category 5, Topic 3 Message 50 Sun Aug 29, 1993 R.KNOP1 [Rob Knop] at 00:45 EDT Aha! JBEE posted that even as I was composing my reply, but yes, that load address is the standard start of BASIC for the 128, so it may be the BASIC program was intended to run on the 128. -Rob ------------ Category 5, Topic 3 Message 51 Sun Aug 29, 1993 C.DRUTIS [Monsoon] at 04:09 EDT Rob, Yes, I've issued a BANK1 command when I've tried to start if from Basic. I've also issued a G10400 (it resides at $400 in Bank 1) from the monitor. It doesn't crash... it simply stalls. In other words nothing happens. The computer doesn't lock up because it responds to a RS/Restore. This is why I'm puzzled. If it couldn't reach the kernal from Bank 1, it would (or should) break and drop to the monitor. I'm beginning to suspect that the kernal and the 128's editor wasn't designed to interact with Bank 1. I was kinda hoping someone had successfully written a program that uses them from Bank 1. Thanks anyway. :) Oh... I want to put it in Bank 1 because of memory constraints in Bank 0. I also don't want to rewrite the entire thing using INDFET and INDSTA and the like. Call it laziness. <g> [Apptest 00324.A] ------------ Category 5, Topic 3 Message 52 Mon Aug 30, 1993 R.KNOP1 [Rob Knop] at 01:08 EDT Monsoon- re: putting code in Bank 1 because of memory constraints, actually, this is a reasonable thing to do. Speed constraints would be another good reason; if you are writing (say) a text editor that keeps its text in BANK 1, it might be advantageous to write short "memory moving" routines in bank 1, which would save several cycles for each lda and sta as compared to INDSTA and INDFET. One other little obvious thing to check: how have you set the MMU register for common memory? If you haven't messed with it, $400 should be fine. Also, are you using this program at all with BASIC? If so, have you moved up the pointers for the bottom of variable memory? Re: 128's OS interacting with Bank 1, in my experience the monitor works beautifully with Bank 1, no problem there. Again, I am not sure about the Kernal. One thing that might be worth trying: setting break points. You can do this manually, by replacing instructions with BRK instructions. Put these varying distances into your code, and if, when you execute the code, the program breaks out to the monitor from one of these, you know that the code is at least getting that far. Another option would be to check out HEXPERT, which is a simple ML debugger that lets you set breakpoints and step through code. I am not sure that Hexpert will work with code in Bank 1, since (again) I haven't tried it, but it's probably worth looking at. -Rob ------------ Category 5, Topic 3 Message 53 Mon Aug 30, 1993 C.DRUTIS [Monsoon] at 04:09 EDT Rob, Yes, yes, no, yes. :) Did that. The monitor resides in RAM-0. I've used Hexpert. I've also set break points. The program actually WORKS in bank 1, but it doesn't seem to interact with some of the Kernal routines (GETIN and BSOUT for ex.). Yet, I HAVE found a kludgy work-around. What I did was move the program down to $1000 in Bank 1 (or, as you say, RAM-1 ), I then shifted Common Ram to 4k. I found a few spare bytes in that 4k and put in three subroutines that did the actual JSRs to the Kernal routines. Then I JSR to those routines from the program in place of the original Kernal calls. Seems to work. BTW, after much searching, I did find a couple of references to RAM 1 not working with the Kernal. On page 414 of the 128-PRG it says that you must be in RAM-0 (with I/O and high ROM). And on page 403 it states that you should invoke configurations 12, 13, 14 or 15; all of which use RAM-0. Thanks for your help. :) [Apptest 00334.A] ------------ Category 5, Topic 3 Message 54 Tue Aug 31, 1993 R.KNOP1 [Rob Knop] at 01:34 EDT One thing you can do is use JSRFAR (or eqivalent) to call the Kernal routines- all it takes is changing one byte in the RAM routine to return to another configuration other than bank 15. Myself, I don't really like JSRFAR (too much of a pain to use), so I brewed up something else that I call "JSROB" which is very much like what you describe doing with your "switchboard" routines in the 4K of common RAM- the main difference being that I put JSROB in zero page (obviously, this won't work if you want to use BASIC with the program). This requires less setup than JSRFAR, saves and restores the registers itself, and returns to the bank from whence it was called. IMHO such an entity is a better and safer way to call Kernal routines from Bank 1 than changing the return configuration of JSRFAR. If anybody is interested I could post the routine here (it's short). Re: terminology, I use RAM-0 and RAM-1 to refer to the RAM blocks in the 128, and Bank n to refer to the 128 standard memory configuration number n. Although bank 0 and bank 1 are (roughly) the same is RAM-0 and RAM-1, this way I avoid confusing myself when I start talking about bank 0 RAM being visible in bank 15 and such things. -Rob ------------ Category 5, Topic 3 Message 55 Thu Sep 02, 1993 T.RUTTER2 [Troy Rutter] at 21:18 EDT Indeed, Modlink thorws the files where they are supposed to be when it runs. As for moving basic up... I do use that trick... but later in the program.