The first disk in the AppleIIAsm Collection includes all of the macros, subroutines, vectors and reserved memory locations required for the rest of the library to function properly. Note that "functioning properly" only applies to unmodified codebases; highly optimized code that makes the library barely recognizable may not use the same standard procedures.
You may consider the files on this disk to provide the overall software architecture for the rest of the library (as outline in the Software Architecture section). However, one of the goals is to be able to remove or alter major components of the architecture--the girders, you might say--in order to lighten the building as a whole while still keeping its basic structure intact. That is, a good deal of effort is done to make this package of libraries as easy for beginners to use as possible while also allowing for substantial changes to the architecture to allow for further optimization after the fact.
Since this will encompass the same group of macros, the information will be contained in a section of its own However, many more macros have yet to be covered, and will be so largely in alphabetical order. There are times, however, when it makes sense to group together subroutines or macros non-alphabetically; this is done on occasion.
The next components are the `_PRN` and `_WAIT` macros, alternatives of which are technically found on other disks but are useful enough to also include in the required library; the subroutines for these will be covered as well..\ However, following that will be a long series of macros that being with "B," which are versions of 6502 branch instructions that allow for longer calls. These are useful enough to include in the required library, and take up little enough space in terms of bytes and cycles that their inclusion here is negligible.
Following the branching macros, two macros dedicated to clear the highest and lowest nibbles of the value in .A are included, following then a series of macros beginning with "C" that emulate some of the functionality found in the 65C02 processor. Additionally, a group of mostly unrelated macros are listed, along with their corresponding subroutines, that are most often used for debugging. Then, finally, 8080 and 780 instruction set macro emulations will be covered briefly, for use by those who have yet to fully internalize the 6502 instruction set.
Without the required header file, almost nothing in the rest of the collection will function properly. This is because certain areas of memory defined in the header file are used by their libraries for holding variable data, referring to zero-page locations, and so on. The listing is as follows, with explanations afterwards (beyond the inline comments):
| Condition | Value |
| --- | --- |
| Name | File: HEAD.REQUIRED |
| Type | Header File |
| Author | Nathan Riggs |
| Last Revision | 29-NOV-2019 |
| Assembler | Merlin 8 Pro |
| OS | Apple DOS 3.3 |
| Purpose | Initialize system for use by rest of libraries |
| Dependencies | none |
| Bytes | 311 |
| Notes | Required for nearly every macro or subroutine |
---
*DETAILS*
First we have the file header, which should be the first section of every file regardless of its type. This includes a short description of what the file contains, last update, contact information, and so on. This file header additionally contains the number of bytes the file uses, and this one in particular uses an unusually substantial number of bytes for a header file (in the optimizations appendix, you'll find ways to reduce the number of bytes here).
From the beginning, the required header file provides space for a small jump table that begins with a jump to the end of the header file, but allows for an additional 16 more addresses for use in the table. Note that you can of course increase or decrease the size of the jump table, but be sure to do so in terms of words (2 bytes), and never get rid of the initial jump to the end of the header file--otherwise, date will be executed as code and an error will occur. Then follows 20 bytes for variable definitions by the rest of the library, titled `VARTAB`. While this can also be increased if the end user also wishes to use space here (though that can be haphazard), reducing the number of bytes here will break certain, more complicated subroutines and thus should not be lowered.
What then follows is 256 bytes dedicated to values returned by subroutines. The first byte, `RETLEN`, is a length byte that determines how long the return value is in `RETURN`. This return length is always initialized upon a return, whether the return value is a single byte or an array. If the value type is a string, then the string length is returned in `RETLEN` while the actual string data is placed in `RETURN`. This is important to remember when referring to strings in loops, as very often the initial instinct is to seek the length of the string at byte zero of `RETURN`.
The rest of the header defines certain hooks and variables that the rest of the libraries depend on. For instance, `ADDR1` through `ADDR4` are often used for indirect addressing modes, and are helpful in the sense that they are all sequential: a high nibble precedes a low nibble, for instance, in **$06** and **$07**. Following the addressing indices, scratchpad locations on the zero page are identified as `SCRATCH` and `SCRATCH2`, though `SCRATCH` is most often used for backing up **.Y** values before and after a macro and thus should rarely be used beyond that capacity. The last variable, `RETVAL`, is used on the rare occasion when usage of the stack requires a return location to be backed up elsewhere.
The hooks `RESULT` and `RESULT2` are rarely used, though this may change in the future. The following word and byte parameter passing hooks, however, are likely the most used parts of the zero page in this collection: `WPAR1`, `WPAR2`,`WPAR3` and `WPAR4` are all used primarily for passing word-length values (16-bits) to subroutines, and have sequential memory locations, whereas `BPAR1`, `BPAR2`, `BPAR3` and `BPAR4` are used for passing single bytes (these do not have sequential free memory locations, and are thus unsuitable for anything by bytes) .
Finally, the warm DOS `REENTRY` hook is defined, which is required to be jumped to at the end of each and every program (for all intents and purposes, anyhow). A few hooks are defined for memory operations, and then we have the `MAIN START` label, which signals the beginning of the actual code of the program that follow.
`Listing 1.1: HEAD.REQUIRED CODE`
```asm
*
*``````````````````````````````*
* HEAD.REQUIRED *
* *
* THIS HEADER MUST BE THE *
* INCLUDED BEFORE ANY OTHER *
* CODE IN ORDER FOR THE PROPER *
* FUNCTIONING OF ANY LIBRARY. *
* *
* AUTHOR: NATHAN RIGGS *
* CONTACT: NATHAN.RIGGS@ *
* OUTLOOK.COM *
* *
* DATE: 11-DEC-2019 *
* ASSEMBLER: MERLIN 8 PRO *
* OS: DOS 3.3 *
* *
* BYTES: 311 *
*,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,*
*
* VARIABLE DECLARATIONS ********
*
** JUMP TABLE SETUP. THIS IS FOR LOADING
** SUBROUTINES INTO MEMORY FOR ACCESS BY
** EXTERNAL EXECUTIONS. NOTE THAT THIS
** SHOULD ALWAYS START AT SECOND BYTE OF
** CODE IN THE PROGRAM SO THAT ITS
** LOCATION IN MEMORY IS EASILY KNOWN.
*
JUMPTBL JMP MAIN_START ; ** ALWAYS ** START WITH
; JUMP TO MAIN_START
DS 32 ; 16 MORE ENTRIES
*
** 20 BYTES FOR VARIABLES
*
VARTAB DS 20
*
** 256 BYTES DEDICATED TO RETURN
** VALUES OF VARIABLE LENGTH; CAN BE
** MODIFIED TO SUIT SMALLER OR LARGER
** NEEDS.
*
RETLEN DS 1 ; RETURN VALUE BYTE LENGTH
RETURN DS 255
*
** ADDRESS STORAGE LOCATIONS FOR
** INDIRECT ADDRESSING.
*
ADDR1 EQU $06 ; AND $07
ADDR2 EQU $08 ; AND $09
ADDR3 EQU $EB ; AND $EC
ADDR4 EQU $ED ; AND $EE
*
** SCRATCHPAD ZERO PAGE LOCATIONS AND
** DEDICATED ZERO PAGE ADDRESS TO HOLD
** A RETURN ADDRESS PASSED VIA THE STACK
*
SCRATCH EQU $19 ; USED TO BACKUP .Y
SCRATCH2 EQU $1E
RETADR EQU $FE ; AND $FF
*
** ZERO PAGE ADDRESSES DEDICATED TO PASSING
** BACK RESULTS WHEN THERE ARE MORE THAN
** THREE BYTES BEING PASSED (AXY) AND THE
** USE OF THE STACK IS IMPRACTICAL OR TOO SLOW
*
RESULT EQU $FA
RESULT2 EQU $FC
*
** WORD AND BYTE PARAMETER SPACE USED
** BY APPLEIIASM MACROS
*
WPAR1 EQU $FA
WPAR2 EQU $FC
WPAR3 EQU $FE
BPAR1 EQU $EF
BPAR2 EQU $E3
BPAR3 EQU $1E
BPAR4 EQU $19
*
** VARIOUS HOOKS USED BY ALL ROUTINES
*
REENTRY EQU $3D0
*
PROMPT EQU $33 ; DOS PROMPT CHARACTER
COLDENT EQU $03D3 ; COLD ENTRY TO DOS
SRESET EQU $03F2 ; SOFT RESET
PRNTAX EQU $F941 ; PRINT HEX VALS OF A,X REGISTERS
BELL EQU $FBE4 ; RING MY BELL
IOSAVE EQU $FF4A ; SAVE CURRENT STATE OF REGISTERS
As stated in earlier sections of this text, each file contains its own heading with information about the content therein; the required macros file is no different. As will all libraries in AppleIIAsm, we will begin with macros because they are the most used (and most useful) way to interface with the collection of libraries. If a macro uses a particular subroutine, that subroutine will be discussed in relationship to the macro rather than be introduced as a separate entity. First, however, we'll list the macro file Header to stress again how such headers will be presented in the future.
A typical macro file header includes the name of the file at the top, followed by a short paragraph that describes what these macros are for. Then, the (main) author and his or her contact information is listed, along with the last date of revision, the Operating system targeted, and the Assembler used. Unless this project grow far beyond my original intentions, these will all likely remain the same, save for revision dates.
What follows then is a listing of all subroutine files that are used by the macros in the file; luckily, there is only one subroutine file in use for the required library. After that, a listing of macros available in the file are listed, with extremely brief descriptions of what they are for; more detailed descriptions can be found in the headers of the macros themselves. A bit strangely, this macro file header contains fewer subroutine listings that most others and more macro listing than most other file headers, as you will see.
Also note that this header carries with it some key variables that are used in the rest of the library. This is usually relegated to the HEAD file of a library, but assembly variables, rather than vectors, are often defined within the file they are used.
The first six macros listed in the required library are primarily for the software architecture's use only, although some may find limited use for them outside of the collection itself. As macros, these are likely the most called---and thus most important, in terms of optimization---among all other macros, and deserve special mention as a group, together. Ultimately, these macros are used to determine how parameters are passed to different subroutines, with the associated byte and cycle costs differing widely depending on which macro and method is used. Among any other reasons, then, this is why they happen to be the most used macros.
Yet what the macros _really_ do is still up for grabs. Before describing each on its own, it helps to remember that the macro names are mnemonic: `_AXLIT` passes variables via **.A** and **.X** while checking whether the variable sent is a literal value or not, `_AXSTR` passes variables (or more correctly, pointers) via **.A** and **.X** while determining if the initial parameter is a string or address; `_ISLIT` checks whether a parameter is a literal and passes the correct value via the stack, whereas `_ISSTR` does the same with strings, and `_MLIT` and `_MSTR` do much the same as the previous macros, except pass values to subroutines via zero page memory dedicated to parameter passing: `WPAR1`, `BPAR1`, and so on.
While these all behave much the same, there are some important differences worth noting, thus good reasoning for tearing apart each macro line-by-line. Note that a subroutine always expects a particular macro to pass parameters; if the wrong macro is used, the subroutine will either freeze or return an error. Instead of complete consistency in terms of passing data to subroutines, we have instead opted for efficiency.
---
### THE \_AXLIT MACRO
_SUMMARY_
| Condition | Value |
| --- | --- |
| Name | `_AXLIT` |
| Type | Macro |
| Author | Nathan Riggs |
| Last Revision | 29-NOV-2019 |
| Assembler | Merlin Pro 8 |
| OS | Apple DOS 3.3 |
| Purpose | parse macro parameters for subroutine passing |
Let us first look at the `_AXLIT` macro, as it serves as the first macro alphabetically as well as succinctly illustrates how the other macros work as well. But first, just like with everything else, we should start with the heading.
At first, you should notice that each and every macro (and subroutine, for that matter) also includes its own heading, like each file. These headings differ in focus, however: while the name and short description remain the same, there is also a listing of parameters that are passed to the macro as well as the number of cycles and bytes a macro will use every time it is called. Note that this is slightly different than a subroutine; since a subroutine is listed only once in memory, it only uses the number of bytes listed once, and does not add to the byte count with any further usage. This is primarily why subroutines and macros exist separately: sometimes it costs more to `JSR` to a subroutine and then `RTS` back to the main listing than it would take to simply write out the entire subroutine and add more bytes. There is far more to it than this, of course, but this is a good foundation for understanding a key difference, especially in terms of optimization.
Note that `_AXLIT` expects a single parameter, denoted by **]1**, and will *at least* use up 8 cycles and 6 bytes per usage. Note that I tend to take the pessimists side, as you'll soon see, and would rather add possible cycles or bytes rather than assume fewer of each. One runs into the problem of execution being too slow far more often than one does too fast.
Now let's look at the `_AXLIT` listing itself, explaining both what each line accomplishes as well as some of the more peculiar notations used in the inline comments.
Warning: If you are unfamiliar with how Assembly language is formatted, this is not the best place to learn so; please see an introductory guide before attempting to read this document any further. Your best bet would be to start with the first few chapters of the first book in this technical documentation series, _A New Users' Guide to Merlin 8 Pro_, to familiarize yourself with the interface, then move to the next book,_From AppleSoft to AppleChop to ASM_ to move from a BASIC mindset (or Pascal, etc.) to an Assembly mindset. Of course, you are more than welcome to use other resources instead.
This macro, of course, begins by declaring `_AXLIT` as a macro on the first line \(the lines are not numbered, so be prepared for some confusion with longer macro and subroutine listings\). The second line uses what is called a "Pseudo Opcode" provided by the Merlin Pro 8 Assembler to test whether the first character of the first \(and only\) parameter is a pound sign, which denotes whether the value being sent is a literal value or an address. If this happens to be a literal, then the next line divides the byte's high nibble of the parameter by $100 and stores it in **.X**, then loads the low nibble in **.A**.
But why? The reasoning is simple enough: when a literal value is passed as a parameter, it is still treated as an address, but one that indirectly points to another address. This provides all memory locations with the capability of zero-page addresses indirectly pointing to an address. This does not mean that these memory locations can be used as indirect addresses in the source code, but it does mean that user-inputted data (or file-inputted data) can be addressed indirectly without overburdening the zero page. Thus, when "#" is detected as the first character, it is assumed that the actual address being passed is stored in the literal value converted into an address.
If the first character is not a pound sign, however, then the parameter is assumed to be a direct address, and is treated as such: the `ELSE` statement tells the assembler to load **.X** with the passed address **+1**, indicating its higher byte, whereas **.A** then stores the low byte of the address to be used. The `FIN` statement acts like a high-level "end if" statement, and the `<<<` opcode represents the end of the macro. Control is then returned to the line following that which called the macro in the first place which usually JSRs a subroutine that expects address data passed in **.A** (low byte) and **.X** (high byte).
Rather quickly, you may notice some strange notation at the beginning of some lines the inline comments, like the third line's _{4C3B}_. This notation documents how many cycles and bytes that the line has added to the total (four cycles, three bytes), and the notation allows for the code minifier utility to skip or include this information in the minified version of the macro or subroutine. This notation is also used at the end of a comment to let the user know which status flags are affected. Again, note that in terms of bytes, this means something different depending on whether it is found in a macro or a subroutine: in a macro, the bytes are always added at each call, whereas in a subroutine the bytes are only added once.
There are other inline comment notations that will be encountered later on---most notably, the double-minus, which says to keep the entire line during minification---but we will describe those as they are encountered. A full list of these meta-notations can be found in Appendix XXX.
While the `_AXSTR` macro works much the same as `_AXLIT`, there are some important differences between testing for string parameters versus literal values, and the `_AXSTR` macro will even call `_AXLIT` if the parameter turns out not to be a literal string. Some subroutines (or rather their calling macros) can accept literal string values to make Assembly more "user-friendly" like higher level languages. One primary example of the is printing to the screen: with the \_PRN macro, we can write `_PRN "hello, world!"` and the macro will act as expected: "hello, world!" will be printed to the screen at wherever the current cursor position resides.
One problem with using the `_AXSTR` macro, however, is that bytes will accumulate rather quickly, turning what should be 1K program into 10K-15K in no time. This is because of the way that the `_AXSTR` macro works, as well as with `_ISSTR` and `_MSTR`. But to understand how this works, let's first look again at the macro header:
For the most part, there is nothing to worry about here: this is a standard macro header. However, note that in the BYTES section, you'll see "9 + STRING BYTES." This means that for each time `_AXSTR` is called with a literal string as a parameter, that string is declared somewhere to become part of the byte-length of the total! A 9 byte macro, then, can easily become a 100 byte macro, and when used over and over again, those bytes can add up. Let's see how this works via the source in question:
As usual, the first line declares `_AXSTR` to be a macro. In this macro, however, the second string's pseudo-operation tests whether the first character of the first and only parameter is a quotation mark that delimits a string; note that for using other text delimiters, which is perfectly acceptable in Merlin Pro 8, you would have to duplicate this code to search for those characters as well. The next line is where the macro starts to significantly differ from `_AXLIT`: if indeed the parameter is a literal string, then a `JMP` command is initiation to bypass the declaration of a temporary string so that the string data itself is not executed. Then, **.X** is loaded with the high byte of this string's address, while the low byte is held in **.A**.
An alternative way to do this, which might be implemented in the near future, is to simply have pre-allocated space for these strings declared by the required header, thus adding 255 bytes and nothing more regardless of how may times the `_AXSTR` macro is used. This, however, will be a practice in trading bytes for cycles, and such a macro will better belong in the STDIO library.
Beyond that, if the first character is not a quotation mark, the parameter is then tested by `_AXLIT` to determine whether it is literal or indirect address being passed to the macro. The `IF-ELSE` opcode is then ended with `FIN` and the macro is ended with `<<<`, returning control to main execution and likely running the appropriate subroutine.
`_ISLIT`, like `_AXLIT` before it, checks if a parameter is a literal or indirect address pass, then parses the address appropriately for an awaiting subroutine. At this point, it would be fairly pointless to separate the header from the rest of the parameter-parsing routines, as the macros here all behave alike, so we will now list them together before dissecting the code.
`_ISLIT` differs from `_AXLIT` primarily in terms of how the address parameters \(pointers\) are passed to a subroutine called after the macro is processed. Instead of passing data via the registers, `_ISLIT` passes data via the stack. Thus, the first line again establishes `_ISLIT` as a macro, while the second line tests whether the first and only parameter is a literal value. If the value is literal, it is first divided by **$100** to get the high byte and then pushes it to the stack, then pushes the low byte to the stack. Otherwise, nearly the same process is made for a nonliteral, except the high byte is literally one byte address above the address passed already; thus, **]1+1**, the high byte, is pushed to the stack, followed then by the low byte. These are then read by a following subroutine backwards: low byte first, high byte second. Confusing this process can lead to countless headaches, so be sure to follow them precisely. On the upside, however, passing via the stack allows for many more variables to be passed to a subroutine than either `_AXLIT` / `_AXSTR` or the soon to be introduced zero-page memory massing routines.
Like before, `FIN` acts as an `END-IF` while `<<<` indicates the end of the macro. Such instructions will not be explained again for the rest of this section on parameter-passing macros.
The `_ISSTR` macro once again works like the `_AXSTR` macro, except that again the parameter pointers are passed via the stack rather than the registers. The first parameter is checked whether it is a literal string; if so, the string is then created on-the-fly and a pointer to that address is pushed to the stack, high byte first then low byte. Otherwise, the parameter is assumed to be an address and is again checked to be a literal; if so, the address is sent to `_ISLIT` for further parsing.
The `_MLIT` macro passes address pointer parameters to a subroutine via predetermined locations on the zero page---those defined by `WPAR1` through `WPAR4` for 16-bit values and those between `BPAR1` through `BPAR4` for single-byte values. Unlike previous macros, `_MLIT` now accepts and introduces the concept of two parameters, defined by the symbols **]1** for the first parameter and **]2** for the second. The first parameter should already be familiar: it's the pointer to the address being passed, and may be either a literal value or otherwise. The second parameter, however, is a bit more complicated due to the fact that the address passed there **must** match the address expected by the subroutine that follows the macro. This is why the hooks `WPAR1` or `BPAR` are provided in the first place: to provide an easy-to-remember method of passing a variable via zero-page addresses.
Like with the order of passing parameters on the stack, it is extremely important that the same memory locations be used in sending the parameters that the subroutine will be expecting--otherwise, it will be working with the incorrect data, leading to errors or a frozen computer. To pass more than a single parameter, the a semicolon is used to separate the values, as such:
`_MLIT #$300;WPAR1`
This passes the contents found at the address held in **$300** via `WPAR1` on the zero page, which corresponds with bytes **$06** and **$07**. By using multiple `_MLIT` macro calls, one can send a total of 12 bytes worth of parameters via the zero page, which is *almost* as must as most anything would need. If one needs more than that, they would use the `_ISLIT` macro.
`_MLIT` mostly works like the parameter-passing macros before it. The first parameter is checked as to whether it is a literal or not; if it is, then the address is divided by **$100** to retrieve the high byte of the address and this is stored in the provided variable location's high byte (say, `_MLIT1`). The low byte is then copied to the low byte are of `_MLIT1.` If the parameter is not a literal value, then the first parameter is copied to the second parameter, both low byte and high byte, in a straight-forward manner.
Lastly, the `_MSTR` macro is used to pass a string or string address to a subroutine called after invoking the macro, like the other string parameter routines. The difference between this routine and the other string subroutines, of course, is that this uses a second parameter to indicate where on the zero page the address to the appropriate data is stored.
The first parameter is checked firsthand for a preceding quotation mark, signaling that it is a literal string; if so, the string is stored in memory and the pointer to that particular address is then stored in the second parameter address, byte-for-byte. If the first parameter is not a string, this the parameter is then checked by `_MLIT` and treated appropriately before being finally prepared for passage to a subroutine.
There are times when you will either want to quickly print some text to the screen or wait for the user to press a key before continuing without importing the macros from the STDIO library. This is especially useful when debugging and writing demo programs for libraries separate from STDIO, as you will see. To accommodate this need, the `_PRN` and the `_WAIT` macros have been included in the required library. Note, however, that these are severely limited macros and corresponding subroutines; when anything remotely complicated or versatile is necessary, it is advisable to use the STDIO library.
The `_PRN` macro is a quick and dirty method of printing to the screen via COUT. Primarily, this macro is used when the STDIO library is not being used but some output needs to be put onto the screen, especially in terms of debugging. It accepts a single parameter that is either a literal string---nothing more, nothing less. Note that this macro, as you'll see in the source, can quickly add bytes to your program's total byte size. Let's skip the `_PRN` macro's header to be closer to the source we are interpreting.
As usual, the first line of the macro defines its name; in this case, `_PRN`. The second line backs up the value in .Y as a courtesy to the programmer of the calling routine, since .Y is often used as a looping counter \(most all subroutines follow this same example, though there are a few exceptions\). The third line, however, may come as a surprise: the subroutine `__P` is called, despite apparently no preparations for passing it a string parameter. While this will become more apparent once the `__P` subroutine is discussed, for now it suffices to say that what is passed to the `__P` subroutine is the current program counter itself, as it is pushed to the stack by the JSR opcode.
After the subroutine is called, somewhat confusingly at first, an ASC is declared with the string passed as a parameter to the `_PRN` macro, followed by a HEX byte of $00. **.Y** is then loaded back with its previously backed up value, and control is returned to the main routine.
Of course, the big question is: how does `__P` know what to print? Those of you fairly well-versed will immediately recognize the trick---and a dirty trick it is. Let's look at the source code for the `__P` subroutine to see exactly how this works. In the meantime, this will also be the first subroutine examined of all subroutines in the collection, so we may have to explain a bit more than the subroutine would usually merit.
The source is as follows. First and foremost, of course, we have the subroutine header. This differs slightly from the header of a macro in that it specifies methods of input and output rather than feign any parameters, and the header details any of the status registers it alters \(or, in the lingo of the header, "destroys"\). Again also note that since a subroutine is placed into memory only once, unlike a macro, the number of bytes it uses does not increase or multiply on each use.
### THE \_\_P SUBROUTINE
_SUMMARY_
| Condition | Value |
| --- | --- |
| Name | `__P` |
| Type | Subroutine |
| Author | Nathan Riggs |
| Last Revision | 29-NOV-2019 |
| Assembler | Merlin 8 Pro |
| OS | Apple DOS 3.3 |
| Purpose | |
| Input | `ASC` string trailing call to subroutine |
As the INPUT section of the header reveals, this subroutine reads the lines _after_ the call to the `__P` subroutine---the question is, how? With a little work, we can explain how that happens here.
The first line, `__P`, declares the start of the subroutine. Afterwards, the first thing that happens is a `PLA` opcode, which pulls the top value off of the stack and stores it in **.A**. This is then stored in the low byte of `ADDR1`, which is, if you don't recall, a 16-bit group on the zero page meant to hold addresses that will be accessed indirectly. Next another byte is pulled from the stack and stored in **.A**, then transferred to `ADDR1` + 1; that is, it is the high byte of `ADDR1`. Essentially, what has just been pulled from the stack is the address of the instruction that called this subroutine from the macro.
Next, a **.Y** offset of 1 is established for indirectly accessing the contents off the calling address+1 byte---that is, the ASC data following the call to `__P` in the macro! This line just so happens to begin with the `:LP` label so that (As per Merlin's rules, the preceding colon means that it is a local label) .**.Y** can be increased and then the next byte in the string printed. This is repeated until the `$00 HEX` byte is encountered, at which point the entire string has been printed to the screen.
The loop is then ended, as `BNE :LP` indicates that as long as .A is not holding $00, the loop should continue. We have hit the $00, so we now continue to the next line, labeled `:DONE`, where we clear the carry bit. We now transfer the offset value stored in **.Y** to **.A**, and add that to the original address we pulled from the stack, effectively skipping the string and `#HEX $00` to the next address pointer. This address is then pushed to the stack and JSR is called, an opcode that jumps to the address at the top of the stack.
We have now officially left the subroutine, and returned to the macro---except now the program counter points at the line after `HEX $00`, the `LDY SCRATCH` line, which restores the original **.Y** value before the macro was called. The macro is then ended with `<<<`, and our first subroutine adventure is complete!
Thankfully, the next subroutine we'll be exploring is much simpler, as well as the macro that calls it. The `_WAIT` macro simply calls the `__W` subroutine, which then loops until a key is pressed on the keyboard or gamepad.
The last of the quick and dirty macros for error testing is the `BEEP` macro, which turns out to be a rather self-explanatory title. This macro sends a click to the internal speaker for the specified number of cycles; the longer this lasts, the more annoying it gets. The macro stands alone and does not rely on any subroutines, due to its inherent simplicity.
Long branches are branching macros that work pretty much exactly like the branching mechanism they represent, except they can jump to an address beyond a single page away. This is especially helpful in prototyping, as normal branching routines often start failing once you begin to add content between the branching instruction and the address to branch to. However, without quite a bit of reworking, using these macros also makes your code non-relocatable, as specific addresses are `JMP'd` to that may no longer exist as first envisioned.
| Notes | Obviously, use this only if BCC won't work |
---
_DETAILS_
The BCCL macro simply performs a BCC instruction and, if the condition is true, then branches to a nearby instruction that performs a JMP to the actual address you want the the BCC to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BCC` instruction stands for _Branch Carry Clear_, and branches execution of a program to a specified address when the carry flag is set to 0. `BCC` is also used to test whether a `CMP` comparison yields a result in which the **.A** register holds a value less than the comparison value. Thus, `BCCL` simply does the same, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BCC won't work |
---
_DETAILS_
The BCSL macro simply performs a BCS instruction and, if the condition is true, then branches to a nearby instruction that performs a JMP to the actual address you want the the BCS to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BCS` instruction stands for _Branch Carry SET_, and branches execution of a program to a specified address when the carry flag is set to 1. `BCS` is also used to test whether a `CMP` comparison yields a result in which the **.A** register holds a value greater than or equal to the comparison value. Thus, `BCSL` simply does the same, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BEQ won't work |
---
_DETAILS_
The BEQL macro simply performs a BEQ instruction and, if the condition is true, then branches to a nearby instruction that performs a JMP to the actual address you want the the BEQ to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BEQ` instruction stands for _Branch Equal_, and branches execution of a program to a specified address when the zero flag is set. `BEQ` is also used to test whether a `CMP` comparison yields a result in which the **.A** register holds a value equal to the comparison value. `BEQ` can also be used to test whether many other operations yield a 0 result, and is often used to test whether a loop should end. Regardless, `BEQL` simply does the same as `BEQ`, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BMI won't work |
---
_DETAILS_
The `BMIL` macro simply performs a `BMI` instruction and, if the condition is true, then branches to a nearby instruction that performs a `JMP` to the actual address you want the the BMI to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BMI` instruction stands for _Branch MINUS_, and branches execution of a program to a specified address when the negative flag is set. `BMI` is obviously used to test if a value is negative, but also doubles as an easy way to test whether the high bit of a value is set. `BMIL` simply does the same as `BMI`, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BNE won't work |
---
_DETAILS_
The `BNEL` macro simply performs a` BNE` instruction and, if the condition is true, then branches to a nearby instruction that performs a `JMP` to the actual address you want the the `BNE` to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BNE` instruction stands for _Branch Not Equal_, and branches execution of a program to a specified address when the zero flag is not set. `BNE` is often used at the end of a loop to determine if the decreasing index has reached zero yet, in which case the loop would be exited. `BNEL` simply does the same, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BPL won't work |
---
_DETAILS_
The `BPLL` macro simply performs a `BPL` instruction and, if the condition is true, then branches to a nearby instruction that performs a `JMP` to the actual address you want the the `BPL` to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BPL` instruction stands for _Branch Plus_, and branches execution of a program to a specified address when the negative flag is not set. `BEQ` is also used to test whether a value has its high bit set; if it is not set, the `BPL` returns nothing. `BPLL` simply does the same as `BPL`, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BVC won't work |
---
_DETAILS_
The `BVCL` macro simply performs a `BVC` instruction and, if the condition is true, then branches to a nearby instruction that performs a `JMP` to the actual address you want the the `BVC` to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BVC` instruction stands for _Branch oVerflow Clear, and branches execution of a program to a specified address when the overflow flag is clear. `BVC` is often used in signed number calculations, but can also be used as a fake `BRA` in relocatable code. Regardless, `BVCL` simply does the same as `BVC`, except is capable of branching to an address that is further away.
| Notes | Obviously, use this only if BVS won't work |
---
_DETAILS_
The `BVSL` macro simply performs a `BVS` instruction and, if the condition is true, then branches to a nearby instruction that performs a `JMP` to the actual address you want the the `BVS` to point to. The purpose of this macro, as well as every **Long Branch** macro is clear: to allow for branching further away than normal. However, not that this should only be used if necessary; otherwise, it becomes a waste of both bits and cycles.
For those of you who need a reminder: the `BVS` instruction stands for _Branch oVerflow Set_, and branches execution of a program to a specified address when the overflow flag is set. `BVS` is often used in signed arithmetic. `BVSL` simply does the same as `BEQ`, except is capable of branching to an address that is further away.
These macros try to mimic some of the functionality of the 65c02 instruction set. For the most part, this revolves around pushing and pulling different registers other than **.A**, with one macro dedicated to setting a memory location to a zero value. These are to be used, of course, only in cases when the actual 65c02 instruction set is unavailable.
`CBRA` stands for "65**_C_**02 Branch Always, a function that is not available in the original 6502 instruction set. All we do here is replace that macro with a JMP opcode..
`CPHX` stands for "65**_C_**02 PusH **.X**, functionality that is not available in the original 6502 instruction set. Due to swapping values in the registers, this does take more cycles than desirable, so when possible just use the `PHA` instruction like every other normal human being.
`CPHY` stands for "65**_C_**02 PusH **.Y**," functionality that is not available in the original 6502 instruction set. Due to swapping values in the registers, this does take more cycles than desirable, so when possible just use the `PHA` instruction like every other normal human being.
`CPLX` stands for "65**_C_**02 Pul**L .X**, functionality that is not available in the original 6502 instruction set. Due to swapping values in the registers, this does take more cycles than desirable, so when possible just use the `PLA` instruction like every other normal human being. Otherwise, this macro pulls the top value from the stack and stores it in **.X**.
`CPLY` stands for "65**_C_**02 PulL **.Y**, functionality that is not available in the original 6502 instruction set. Due to swapping values in the registers, this does take more cycles than desirable, so when possible just use the `PLA` instruction like every other normal human being. Otherwise, this macro pulls the top value from the 6502 stack and stores it in **.Y**.
Sadly, in regular 6502 there is no instruction for transferring the contents of **.X** to **.Y**, like there is in the 65c02 instruction set. This macro, which stands for "65**_C_**02 Transfer **.X** to **.Y**, fakes that functionality by using **.A** as an intermediary and then restoring its original value.
Likewise, in regular 6502 there is no instruction for transferring the contents of **.Y** to **.X**, like there is in the 65c02 instruction set. This macro, which stands for "65**_C_**02 Transfer **.Y** to **.X**, fakes that functionality by using **.A** as an intermediary and then restoring its original value.
While rather sparse, the Required Macros Library does include a number of independent macros that help with debugging, timing, and so on. They are as follows.
The `DELAY` macro calls the `DELAYMS` subroutine following this listing, which holds the 6502 CPU busy for a specified number of milliseconds by repeating through a specific number of cycles. It first backs up the **.Y** register for restoration after the delay, then calls the subroutine.
The `DELAYMS` subroutine uses a precise number cycles to delay for a number of given milliseconds. This is mostly adapted from Leventhal and Seville's _6502 Assembly Language Routines_.
It should be noted that this routine will only work correctly on a 6502 CPU that runs at 1.23 mhz; any other speed will slightly alter the delay.
It is often useful, and sometimes necessary, to view the contents of a block of memory while trying to debug a given subroutine. The `DUMP` macro does exactly that: it dumps a specified block of memory to the screen for the user to see, in hexadecimal, before continuing execution of the program. Note that this does not ceate a pause for the information to be absorbed; the pause must be explicitly stated with something like a `_WAIT` statement. This macro calls the `__DUMP` subroutine, which handles most of the work.
The `__DUMP` subroutine dumps a given address range of hexadecimal values to the screen, often used for debugging. The Actual hexadecimal values are converted to strings before being printed to the screen, given a starting addres at each line.
`ERRH` is a macro that sets the address that would be pointed to in the case of an error-handling issue in Applesoft \(and sometimes in DOS\). This is often used in conjunction with file handline routines, since many of those are to be found Applesoft. First the **.Y** register is backed up for later restoration, then calls the `_AXLIT` macro to pass parameters to the `__ERRH` subroutine. The **.Y** register is then restored to its original value, and control is returned to the main program.
The `__ERRH` subroutine tricks DOS into thinking it is in immediate mode, turns on error handling, then defines the hook to which an error will point to. Control is then returned to the calling routine.
The `GBIT` macro loads **.A** with the value of a given bit in a given byte. To address which bit to test, the BITON# variables should be used in order to avoid "magic numbers." For instance, `GBIT $300;#BITON6` would test whether bit 6 of the byte found in address $300 is either a 0 or 1, which is store in the Accumulator. Remember that the eight bits of a byte start at bit 0 from the right to bit 7 on the left. Therefore, the preceding code would test the second from last bit in the byte.
The `GRET` macro pulls the data held in `RETURN` and stores it in an alternative address. The length to be copied is to be determined by the `RETLEN` byte, which precedes `RETURN` in memory.
The remaining macros and subroutines are dedicated to memory manipulation, which is such a common need in most routines that their inclusion here is deemed necessary. Note that rior to this version of the collection, memory macros and subroutines were considered a library separate from the required library, though they shared the same disk.
### THE PEEK MACRO
_SUMMARY_
| Condition | Value |
| --- | --- |
| Name | `PEEK` |
| Type | Macro |
| Author | Nathan Riggs |
| Last Revision | 12-DEC-2019 |
| Assembler | Merlin 8 Pro |
| OS | Apple DOS 3.3 |
| Purpose | loads a value held at an address in **.A** |
The `MFILL` macro takes a starting address a length byte, and a fill value, and fill a block of memory starting at the address given with the specified length and fill value. First, **.Y** is backed up to be restored after calling the `MEMFILL` subroutine, then the parameters are sorted using the `_MLIT` macro for parameters passed via the zero mage. Another parameter is stored in `BPAR1`, then `MEMFILL` is called to complete the task. Once the subroutine returns control back to the macro, **.Y** is restored to its original value.
The `MEMFILL` subroutine is usualy called by `MFILL` in order to fill block of memory with a given value. Note that this can span more than a single page, and fills whole pages before it continues to fill the pages that are not fully overwritten. Also note that the main algorithm for this was waken from Leventhal & Saville's _6502 Assembly Routines_, as it works very well and there is no need, at this time, to completely reinvent the wheel.
The `MOVB` macro moves an number of blocks of memory from a source address to a destination address with a given length. The macro first backs up the **.Y** register, then runs a series of `_MLIT` macros to prepare for the passing of parameters to the `MEMMOVE` subroutine via the zero page. The subroutine is then called, after which the **.Y** register is restored to its original value.
This is another subroutine lifted from Leventhal and Seville's _6502 Assembly Language Routines_. While moving memory contents on its surface does not seem like a terribly complicated process, this implementation makes sure to work with overlaps in copying, etc. in a mostly efficient way. Currently, this author would be able to do no better than Leventhal and Seville here, and thus the subroutine stays. Of course, this may be altered later to a point where Seville & Leventhal's work is no longer required.
Ultimately, the `MEMMOVE` subroutine moves a block of memory at a certain length to another adfress. This block of memory can span multiple pages, and thanks to the way the subroutine functions, overlaps in memory allocation should not be a concern.
The `MSWAP` macro swaps a given block of memory with another block of memory. Note that unlike most other memory routines here, this can only work with up to 255 bytes, and there is no error checking for overlaps. These issues will be addressed in the future, of course, but it would be best to keep this in mind until then.
The macro first backs up the contents of the **.Y** register, then makes several calls to the `_MLIT` macro to sort out parameters on the zero page. **.A** is then loaded with the length in bytes of the swap and stored in `BPAR1` as a parameter before the `MEMSWAP` subroutine is called. Afterwards, control is returned to the main program and the original contents of **.Y** are restored.
The `MEMSWAP` subroutine swaps one block of memory less than a single page to another block of memory. No error checking is applied, especially for cases like memory overlap, so care should be taken with the subroutine as it currently stands.
Disk I also contains alias files that hold macros that emulate parts of the instruction sets of different processors on the 6502, when possible. Obviously, the cases for which this can happen are limited: the 6502 has fewer registers than most other processors, fewer capabilities in some cases \(which is one reason why it was cheap to use\), and limited addressing capacities. Thus far, the following processors have some alias macros that emulate the same behavior on the 6502:
- 8080/8086 family (up to 286)
- z80 Family of Processors
It is advisable to use only one alias family at a time, as some instruction sets share instruction names. Note that instructions related to mathematics are rarely used here, as these are mainly part of the math library already. Only basic instructions like branching are included in the alias files, and thus the source code for each is rather simple. As such, we won't be listing the source here, but providing tables of the instruction set macros with their cycles, bytes used and purpose. If a particular instruction becomes complicated enough to merit explanation, we will list it separately from the rest of the instruction set table.
### 8080 Instruction Set Macro Substitutions
The following instruction set macro replacements for the 8080/8086 line of processors are largely branching instructions, although two others stand out: `ANC` and `SNC`. These stand for _Add No Carry_ and _Subtract No Carry_, respectively, and will be removed from here and added to the math library once revision once again reaches that disk.
| **`ANC`** | ADD NO CARRY \(ADD in 8080\) | 8 | 12 |
| **`SNC`** | SUBTRACT NO CARRY \(SUB in 8080\) | 8 | 12 |
| `PUSHA` | PUSH ALL REGISTERS | 14 | 30 |
| `PULLA` | PULL ALL REGISTERS | 24 | 19 |
| `POPA` | PULL ALL REGISTERS | 24 | 19 |
### Z80 Family Instruction Set Macro Substitutions
Currently, there are few instruction set macro substitutions for the z80 processor because they operate in very different ways. However, they is some overlap, and as much overlap as possible will be provided by these aliases.
At the the end of each library, a demo is provided that shows the macros being used as they should be (or could be). These demos are not meant to be exhaustive, but are meant to merely illustrate how the macro is called, what parameters it might require, and so on. By and large, the demos are all kept fairly simple due to the fact that 1) they cannot interact with the macros and subroutines provided by other libraries, and 2) they are meant for beginners to see how a macro works and not much else. More complicated demos are planned for future disks in the package, and some of them are already finished (though in need of updating to the current version).
The demo file listed here is the same `DEMO.REQUIRED` file on the disk. Note that most of the heavier descriptions of the macros are done in commenting in order to cut down on the demo file size in bytes, but `_PRN` and `_WAIT` are particularly used in demo files to give some context when the demo itself is executed. Note also that demo files, like all other files, have unique headings before the code proper begins.
