ACME ...the ACME Crossassembler for Multiple Environments --- Quick reference --- This file should give you a basic overview. More specialized stuff like forcing a specific addressing mode is discussed in extra files ("AddrModes.txt" in this case). ---------------------------------------------------------------------- Section: Example of what an ACME source code file looks like ---------------------------------------------------------------------- ;--- Example code fragment, start --- !to "tiny.o", cbm ; set output file and format * = $c000 ; set program counter CLEAR = 147 ; a global symbol definition !addr basout = $ffd2 ; another one, marked as an address ; a string output loop: ldx #0 beq + ; enter loop - jsr basout ; output character inx ; advance pointer + lda .string, x ; get character bne - ; check whether last rts .string !pet "Dumb example", 13, 0 ;--- Example code fragment, end --- Here's the same fragment again, now with some additional info: ;--- Example code fragment, start --- !to "tiny.o", cbm ; set output file and format ; This is a pseudo opcode to select the output filename and format. ; This can also be done using the command line options "-o" and "-f", ; respectively. * = $c000 ; set program counter ; This can also be done using the command line option "--setpc". ; some global symbol definitions CLEAR = 147 ; this is a simple constant ; Now "CLEAR" is defined as a global symbol having the value 147. !addr basout = $ffd2 ; this gets marked as an address ; Now "basout" is defined as a global "address" type symbol having the ; value $ffd2. ; The distinction between addresses and non-addresses only ; matters when the type check system gets activated using ; the "-Wtype-mismatch" switch. Then, a line like ; "lda CLEAR" would trigger a type mismatch warning because ; of the missing '#' character. ; a string output loop: ldx #0 beq + ; enter loop ; "+" is an anonymous forward label. Other ones are "++", "+++", etc. ; They can be used like any other symbol, but they always reference ; their *NEXT* definition. This saves having to think of names for ; unimportant labels. As the label's value is not defined yet, ACME ; will need to perform a second pass. - jsr basout ; output character ; "-" is an anonymous backward label. Other ones are "--", "---", etc. ; They can be used like any other symbol, but they always reference ; their *PREVIOUS* definition. This saves having to think of names for ; unimportant labels. In the line above, the value of "-" is set to ; the current program counter. inx ; advance pointer + lda .string,x ; get character ; Here the value of "+" is set to the current program counter. ; ".string" is a local symbol (because its name starts with a '.' ; character), but as its value is not defined yet, ACME will need to ; perform a second pass. bne - ; check whether last ; Here the last definition of the anonymous "-" label is referenced. rts .string !pet "Dumb example", 13, 0 ; Now the value of the local label ".string" is set to the current ; program counter. All label values are defined now, so after having ; done the second pass, the binary will be saved. The "!pet" pseudo ; opcode stores its string argument in PetSCII encoding to memory, ; followed by the given byte values. ;--- Example code fragment, end --- As you can see, pseudo opcodes are prefixed with an exclamation mark. That's non-standard, but: Backwards compatibility is the root of all evil. :) Summary about symbols: There are global symbols (their names starting with a letter or an underscore character). These can be accessed throughout the whole assembly. Then there are local symbols (their names starting with a '.' character). These can only be accessed from inside the macro or zone they were defined in (for more about macros and zones, see the file "AllPOs.txt"). There are also "cheap locals": their names start with an '@'. The area where these can be accessed is limited automatically by the previous and the following global label (cheap locals are "cheap" because you don't have to put in any extra work to limit their range). And then there are anonymous labels (their names being sequences of either '-' or '+' characters). They are also local (bound to their macro/zone), but in addition to that, the "-" labels can only be used for backward references, while the "+" labels can only be used for forward references. In contrast to global and local labels, anonymous labels can not be defined explicitly (as in SYMBOL = VALUE). Each macro call automatically gets its own scope for local symbols. Save the given example source code to a file called "tiny.a" and start acme by typing acme tiny.a ACME will then parse the file and report any errors. An output file will only be generated if there were no errors and if an output filename has been given. After assembly, the example program can be run on a C64 using LOAD "tiny.o", 8, 1 SYS 49152 Note that ACME does not include any routines for transferring data to a C64. Such tools exist on almost every platform, and I didn't want ACME to become bloatware. ---------------------------------------------------------------------- Section: The pseudo opcodes ---------------------------------------------------------------------- A list with information on how to use all the Pseudo Opcodes can be found in the file "AllPOs.txt". Here's just a short overview: !byte !word !24 !32 !fill !align ...for directly placing values into the output file. !zone !symbollist ...for defining the scope of local symbols and saving global symbols. !convtab !pet !raw !scr !scrxor !text ...for converting and outputting strings. !do !endoffile !for !if !ifdef !ifndef !set !while ...for flow control; looping assembly and conditional assembly. !binary !source !to ...for handling input and output files. !pseudopc ...for offset assembly. !initmem *= ...for segment assembly. !macro + ...for defining and calling macros. !cpu !al !as !rl !rs ...for CPU support, especially the 65816 processor. !warn !error !serious ...for generating warnings, errors and serious errors. !address ...to mark symbols as addresses, for the optional type check system. ---------------------------------------------------------------------- Section: Command line arguments ---------------------------------------------------------------------- The command line syntax for calling acme is quite simple: acme [options] [files] Available options are: -h, --help show this help and exit This is more or less useless, because the help is also shown if ACME is run without any arguments at all. -f, --format FORMAT set output file format Use this with a bogus format type to get a list of all supported ones (as of writing: "plain", "cbm" and "apple") -o, --outfile FILE set output file name Output file name and format can also be given using the "!to" pseudo opcode. If the format is not specified, "!to" defaults to "cbm", while the command line option defaults to "plain". -r, --report set report file name This creates a text listing containing the original line number, the resulting memory address, the byte value(s) put there and the original text line from the source file. -l, --symbollist FILE set symbol list file name This can also be given using the "!symbollist"/"!sl" pseudo opcode. The switch was called "--labeldump" in older versions, that name still works, too. --vicelabels FILE set file name for label dump in VICE format The resulting file uses a format suited for the VICE emulator. --setpc NUMBER set program counter This can also be given in the source code using "* = NUMBER". --cpu CPU_TYPE set target processor This can be changed in the source code using the "!cpu" pseudo opcode. Defaults to 6502. Use this with a bogus cpu type to get a list of all supported ones. --initmem NUMBER define 'empty' memory This can also be given using the "!initmem" pseudo opcode. Defaults to zero. --maxerrors NUMBER set number of errors before exiting If not given, defaults to 10. --maxdepth NUMBER set recursion depth for macro calls and !src The default value for this is 64. --ignore-zeroes do not determine number size by leading zeroes Normally, using leading zeroes forces ACME to generate oversized addressing modes, like 3-byte absolute instructions instead of 2-byte zero page instructions. Using this CLI switch disables this behavior. --strict-segments turn segment overlap warnings into errors When changing the program counter, segment overlap warnings may be generated. Using this CLI switch turns those warnings into errors (which is recommended). This strict behavior may become the default in future releases! -vDIGIT set verbosity level Sets how much additional informational output is generated. Higher values mean more output: acme -v0 source.a This is the default: No additional output is generated, ACME will only display warnings and errors. acme -v1 source.a Now the start and end addresses of the generated output file are displayed, along with its size (a CBM-style "load address" is *not* counted). acme -v2 source.a In addition to the "-v1" output, ACME will announce each pass, will show amount and offset of "!binary" loads, and show start and end addresses and size of each segment. acme -v3 source.a In addition to the "-v2" output, ACME will now announce each source file. -DSYMBOL=VALUE define global symbol This option is useful if you build your projects using Makefiles: "-DSYSTEM=64" could build the C64 version while "-DSYSTEM=128" could build the C128 version of the software (using conditional assembly in your source code file). -I PATH/TO/DIR add search path for input files This option allows to add a directory to the search list for input files. If an input file cannot be found in the current working directory, all directories in the search list are tried (the first match is used). -W fine-tune amount and type of warnings -Wno-label-indent Disables warnings about labels not being in the leftmost column. -Wno-old-for Disables warnings about the old "!for" syntax and at the same time enables warnings about the _new_ "!for" syntax. Internally, this does exactly the same as what happens when the "--dialect 0.94.8" CLI switch is used... -Wno-bin-len Do not complain about unusual number of digits in a binary literal. -Wtype-mismatch Enables type checking system (warns about wrong types). --use-stdout fix for 'Relaunch64' IDE With this option, errors are written to the standard output stream instead of to the standard error stream. --msvc output errors in MS VS format This changes the format of the error output to that used by a certain commercial IDE. --color uses ANSI color codes for error output If your terminal emulation supports ANSI escape codes, use this option to have warnings and errors displayed in color. --fullstop use '.' as pseudo opcode prefix This changes the prefix character used to mark pseudo opcodes from '!' to '.' (so sources intended for other assemblers can be converted with less effort). --dialect VERSION behave like different version This CLI switch tells ACME to mimic the behavior of an older version. Use this with a bogus version to get a list of all supported ones. --test enable experimental features This is for people who want to help test new features before they are officially announced. -V, --version show version and exit. Platform-specific versions of ACME might offer more options. Since version 0.89, ACME accepts more than one top-level-filename given on the command line. ---------------------------------------------------------------------- Section: The expression parser ---------------------------------------------------------------------- ACME has a relatively powerful maths parser. This parser is used whenever ACME expects to read a value. Supported operations include addition, subtraction, multiplication, divisions, comparisons, shifts, negation, boolean operations and some assembler-specific stuff like extracting the "low byte", the "high byte" or the "bank byte" of a value. Calculations are done using either signed (at least 32-bit) integer arithmetic or floating point arithmetic using the C "double" data type. Symbol values are stored the same way. This is a list of the value formats currently known by ACME: Examples Notes --------------------------------------------------------------------- 128 a decimal value, integer 128.5 a decimal value, floating point $d011 hexadecimal values are indicated by either a 0xffd2 leading "$" or a leading "0x". &1701 an octal value, indicated by "&" %1010 binary values are indicated by either a leading "%" %....#... or a leading "0b". In binary values, you can 0b01100110 substitute the characters "0" and "1" by "." and "#" respectively. This way the values are much more readable, especially when building bitmapped objects (like C64 sprites or fonts) in your source code. 'p' single characters in single quotes are converted to their character code. The actual numeric value depends on the current conversion table chosen using the "!ct" pseudo opcode. "player 2" double quotes indicate text strings. See below for more information on single vs. double quotes. [2, 3, 5, 7] brackets indicate lists. These are useful to group [0, [x, y], 9] data, for example when passing an arbitrary number of arguments to a macro. poll_joy2 a global symbol .fail a local symbol, indicated by leading "." @loop a "cheap local", indicated by leading "@" * the current program counter. During offset assembly, "*" gives the value of the "Pseudo PC". Just to make sure: The value of the program counter is always the value that was valid at the start of the current statement, so !word *, *, *, * will give the same value four times. I think most assemblers do it this way. In older versions of ACME, 'x' and "x" were the same thing, namely the character code of the letter x using the currently selected encoding table. Since release 0.97, anything in single quotes gives the character code (as before), while anything in double quotes is treated as a string object. To be compatible to those older versions, ACME keeps accepting one-char strings in a lot of places where actually single characters are expected. This is a list of the operators currently known by ACME: Priority Example Meaning Alias Note ---------------------------------------------------------------------- 16 is_number(v) these three functions return 1 *3 16 is_list(v) if v is the correct symbol *3 16 is_string(v) type and 0 otherwise *3 16 len(v) length of list or string *2 16 sin(v) trigonometric sine function 16 cos(v) trigonometric cosine function 16 tan(v) trigonometric tangent function 16 arcsin(v) inverse of sin() 16 arccos(v) inverse of cos() 16 arctan(v) inverse of tan() 16 address(v) mark as address addr(v) 16 int(v) convert to integer 16 float(v) convert to float 15 &symbol "unpseudopc" symbol (see docs on "!pseudopc") 14 v[w] access v with index w *2 13 ! v bit-wise complement NOT 12 v ^ w to the power of 11 - v negate 10 v * w multiply 10 v / w divide 10 v DIV w integer divide 10 v % w remainder of DIV MOD 9 v + w add *3 9 v - w subtract 8 v << w shift left ASL, LSL 8 v >> w arithmetic shift right ASR 8 v >>> w logical shift right LSR 7 < v low byte of 7 > v high byte of 7 ^ v bank byte of 6 v <= w lower or equal 6 v < w lower than 6 v >= w higher or equal 6 v > w higher than 5 v != w not equal <>, >< *3 4 v = w equal *3 3 v & w bit-wise AND AND 2 bit-wise exclusive OR XOR 1 v | w bit-wise OR OR Notes: "*2" means this operator only works on lists and strings. "*3" means this operator works on all three data types (numbers, lists and strings). All other operators only work on numbers. Operations with higher priority are done first. Of course you can change this using parentheses. Note that though there are operators to extract the "low byte", the "high byte" and the "bank byte", there is no operator to extract the fourth byte. If you want to access that, shift it down using ">>>" or "LSR". In cases where it's not clear which operator was wanted, ACME takes the longest possible one: v<>w ...checks for "v not equal w" v< >w ...checks for "v smaller than high byte of w" So you may have to separate operators with spaces to make sure ACME does what you want. The "power-of" operator is right-associative, so a^b^c means a^(b^c). Calculating 0^0 (zero to the power of zero) will give 1. If you don't know why I'm telling you this, ask a mathematician. :) ---------------------------------------------------------------------- Section: Almost, but not quite, complete syntax ---------------------------------------------------------------------- Every ACME source code file consists of a non-negative number of "lines". The lines have to be separated from each other using CR, LF or CRLF characters. Every line consists of a non-negative number of "statements" and an optional comment. Statements have to be separated from each other using colon (":") characters, the comment has to be prefixed with a semicolon (";") character or two slashes ("//"). Every statement consists of an optional "label" and an optional "command". These are separated from each other using any number of SPACE or TAB characters. If a label has blanks before it, a warning is issued (to spot typing errors - see Errors.txt for more info). Every symbol name consists of these characters: "a" to "z", "A" to "Z", "0" to "9", the underscore character "_" and all characters with values beyond 127. The first character must not be a digit though. But it can be '.' or '@', making the symbol a local one. Local symbols beginning with '.' are only valid inside the current zone (marked using the "!zone" pseudo opcode) or the current macro. Local symbols beginning with '@' are only valid between the enclosing global labels (or inside the current macro). Two other possibilities for label names are "all-characters-are-minus" (then it is an anonymous backward label) and "all-characters-are-plus" (then it is an anonymous forward label). Every command is one of the following: An assembler mnemonic with an optional argument A pseudo opcode, beginning with a "!" character An explicit symbol definition (SYMBOL = VALUE) A pc definition, beginning with a "*" character A macro call, beginning with a "+" character ...and the syntax of those things varies. :) Assembler mnemonics and pseudo opcodes are case insensitive, so whether you write "LDA" or "lda" or "LdA" does not make a difference. In earlier releases of ACME, arithmetic operators like MOD, XOR, LSL had to be written in UPPER CASE. This is no longer needed. Symbol names are case sensitive, so "label" and "Label" are two different things.