64tass Activity
An assembler for the legendary 6502 processor and it's derivatives
Brought to you by:
soci
Activity for 64tass
-
Janne Peräaho posted a comment on ticket #24
Aha, that's the way to do it. Thanks! A little more complicated than in C. PHP has an include_once statement that is really handy and easy to use.
-
Soci/Singular posted a comment on ticket #24
Such double compilation guarding can be done this way: myfile_h :?= false ; define to false if variable doesn't exists yet .if !myfile_h ; was not compiled yet? myfile_h := true ; remember that it is now myconst = 3 .endif This ensures one inclusion per namespace. I guess it is needed to prevent duplicate symbols in a single global namespace, so it should fit.
-
Include only once
-
Only if longer
-
Only at the end
-
Location moves
-
Only at the end
-
Patrick Surry posted a comment on ticket #23
the original use case is to make a header() macro for a Forth. each word is defined by a code block with start_xxx and end_xxx labels. i want a macro like header(xxx) that will generate .word start_xxx, end_xxx as well as .byte len("xxx") and .text "xxx". of course i could pass both string and label args but with above i was able to make it work with just a bare xxx arg.
-
This is because $ is an empty bit string. Similarly like "" is an empty character string or x'' an empty byte string. The numeric value of all three are zero in expressions which need numeric values. While it works as designed I see how an empty (bit) string is unexpected as an instruction argument. Or even elsewhere, as it's basically "nothing". Will think about it.
-
In r3122 the possibility to convert a string to a symbol was added. This is accessible through type conversion (as expected) however there was no built-in name "symbol" for the symbol type before r3137. If using a version somewhere in between the following will define it: symbol = type(.dummy) With the latest trunk this definition is no longer necessary as it has the "symbol" built-in. Given a string of characters which are allowed in symbols the following holds: symbol("foo") == .foo str(.foo) ==...
-
Add symbol as type name
-
progress! some code browsing and experiments with type and repr helped me started to understand the 64tass type system. if I have code like: foo: nop then foo has code type, and repr(foo) uses its address, but the dot operator (mentioned once in passing in the doc) makes .code a symbol, and str(.foo) gives the literal "foo". I still can't figure out how to go the other way, from the literal to the symbol, but this is enough for what I need. I thought maybe ."foo" or an explicit symbol("foo") but...
-
Convert string literal to symbol and/or symbol name to string?
-
Tom Seddon created ticket #74
64tass will assemble `lda #$` as `lda #$00`
-
Cleanup
-
Fix warning
-
Missing reference
-
Improve message
-
Location only instead of address
-
Struct offset vs code
-
Avoid duplicate message
-
Same for alignpageind
-
Do alignment anyway
-
Use member for .from
-
Zibri created ticket #73
not sure if a bug or a feature request...
-
Daniel Tremblay posted a comment on ticket #22
This is a great solution and is solves my problem . Thank you!
-
Hello! In expressions the value of a label is the program counter address at the place where it got defined. The bank operator just returns bits of 16-23 no matter what's thrown at it and because the address space is small it will be always zero. The image file format may be larger than the address space of the CPU that's why * can be assigned to $10000. In this case the program counter will be the last 16 bits but the image offset is not truncated. I think what you're after is the offset of a label...
-
Bank Operator returns 0 for 6502 processor
-
Report other errors first
-
Looks good - thanks! --Tom
-
Surprising direct page address overflow warning
-
That struct offset (S.a) was incorrectly treated as a zero page location. The offset applied ($3a0) would move it off the zero page but the current CPU can't do that so it got wrapped. Correctly the struct offset is just an offset and then the sum of two offsets is not a memory location and so no warning. Fixed it in r3125. Please check.
-
Struct member is not a zp address (#71)
-
conflicting types for 'memalign'
-
I've renamed it in r3124. Thanks!
-
Rename memalign (#72)
-
Stefan Haubenthal created ticket #72
conflicting types for 'memalign'
-
Surprising direct page address overflow warning
-
Might fail
-
Add conversion
-
Use ** for pow()
-
I've added an .elif alias to cover this sort of typo in 1.59.
-
64tass 1.59.3120
-
64tass released /binaries/64tass-1.59.3120.zip
-
-
1.59.3120
-
Update makefiles
-
Add examples
-
Simplify
-
Not on integer division
-
Update list
-
Wrong type
-
Update dependencies
-
Improve command line file errors
-
Use symbols
-
Simplify
-
Don't skip of course...
-
Was not up-to-date
-
Add --output-exec command line option
-
Silly mistake
-
Move it back
-
Update dependencies
-
Expression for labels-root
-
Better do it separately
-
Makes better sense for bits only
-
Add warning for @b, @w, @l future compatibility
-
Reorder
-
Set size as well
-
Add PGX and PGZ output formats
-
Check for empty name
-
Same here
-
Needs more escaping...
-
Doesn't duplicate that way...
-
Update syntax
-
Add link
-
Typo
-
Cleanup
-
Typo
-
Redundant
-
White space
-
Cleanup
-
Careful with duplicates
-
Update syntax files
-
Add .elif alias (same as .elsif)
-
Cleanup
-
Cleanup
-
Use decimal
-
Missing close
-
Simplify
-
Optimize
-
Alternate for should be from bits
-
Must be smaller for that
-
Redundant
-
Proper message
-
Revert
-
This is more correct
-
Fixed 8 bit
-
Simplify
-
Simplify
-
Handle carefully
1 >