If a DCI string ended with a string delimiter or non-ASCII character
(e.g. a PETSCII char with no ASCII equivalent), the code generator
output the last byte as a hex value. This caused an error because it
was outputting the raw hex value, with the high bit already set, which
the assembler did not expect.
This change corrects the behavior for code generation and on-screen
display, and adds a few samples to the regression test suite.
(see issue #102)
On the 65816, if you say "JSR foo" from bank $12, but "foo" is an
address in bank 0, most assemblers will conclude that you're forming
a 16-bit argument with a 16-bit address and assemble happily. 64tass
halts with an error. Up until v1.55 or so, you could fake it out
by supplying a large offset.
This no longer works. The preferred way to say "no really I mean to
do this" is to append ",k" to the operand. We now do that as needed.
I didn't want to define a new ExpressionMode for 64tass just to
support an operand modifier that should probably never actually get
generated (you can't call across banks with JSR!), so this is
implemented with a quirk and an op flag.
64tass v1.56.2625 is now the default.
(issue #104)
The DCI string format uses character values where the high bit of the
last byte differs from the rest of the string. Usually all the high
bits are clear except on the last byte, but SourceGen generally allows
either polarity.
This gets a little uncertain with single-character strings, because
SourceGen can't auto-detect DCI very effectively. A series of bytes
with the high bit set could be a single high-ASCII string or a series
of single-byte DCI strings.
The motivation for allowing them is C64 PETSCII. While ASCII allows
"high ASCII" as an escape hatch, PETSCII doesn't have that option, so
there's no way to mark the data as a character or a string. We still
want to do a bit of screening, but if the user specifies a non-ASCII
character set and the selected bytes have their high bits set, we
want to just treat the whole set as 1-byte DCI.
Some minor adjustments were needed for a couple of validity checks
that expected longer strings.
This adds some short DCI strings in different character sets to the
char-encoding regression tests.
(for issue #102)
64tass wants to place its output into a 64KB region of memory,
starting at the address "*" is set to, and continuing without
wrapping around the end of the bank. Some files aren't meant to be
handled that way, so we need to generate the output differently.
If the file's output fits nicely, it's considered "loadable", and
is generated in the usual way. If it doesn't, it's treated as
"streamable", and the initial "* = addr" directive is omitted
(leaving "*" at zero), and we go straight to ".logical" directives.
65816 code with an initial address outside bank 0 is treated as
"streamable" whether or not the contents fit nicely in the designated
64K area. This caused a minor change to a few of the 65816 tests.
A new test, 20240-large-overlay, exercises "streamable" by creating
a file with eight overlapping 8KB segments that load at $8000.
While the file as a whole fits in 64KB, it wouldn't if loaded at
the desired start address.
Also, updated the regression test harness to report assembler
failure independently of overall test failure. This makes it easier
to confirm that (say) ACME v0.96.4 still works with the code we
generate, even though it doesn't match the expected output (which
was generated for v0.97).
(problem was raised in issue #98)
The code generator outputs an optional comment specifying which
version of which assembler the code was generated for. This was
handled inconsistently and, for the most part, incorrectly. We now
report the correct version.
Two things changed: (1) string literals can now hold backslash
escapes like "\n"; (2) MVN/MVP operands can now be prefixed with '#'.
The former was a breaking change because any string with "\" must
be changed to "\\". This is now handled by the string operand
formatter.
Also, improved test harness output. Show the assembler versions at
the end, and include assembler failure messages in the collected
output.
A few tweaks:
- Test now requires an ORG on offset +000002, not just a correct
address.
- Suppress on-screen display of the initial ORG directive when
a PRG file is detected. Subtle, but helpful.
- In new project setup, fix initial address for PRG projects that
load at $0000.
- In new project setup, add a "load address" comment to the first line.
Also, fix some out-of-date documentation.
(issue #90)
C64 PRG files are pretty common. Their salient feature is that they
start with a 16-bit value that is used as the load address. The
value is commonly generated by the assembler itself, rather than
explicitly added to the source file.
Not all assemblers know what a PRG file is, and some of them handle
it in ways that are difficult to guarantee in SourceGen. ACME adds
the 16-bit header when the output file name ends in ".prg", cc65
uses a modified config file, 64tass uses a different command-line
option, and Merlin 32 has no idea what they are.
This change adds PRG file detection and handling to the 64tass code
generator. Doing so required making a few changes to the gen/asm
interfaces, because we now need to have the generator pass additional
flags to the assembler, and sometimes we need code generation to
start somewhere other than offset zero. Overall the changes were
pretty minor.
The 20042-address-changes test needed a 6502-only variant. A new test
(20040-address-changes) has been added and given a PRG header. As
part of this change the 65816 variant was changed to use addresses
in bank 2, which uncovered a code generation bug that this change
also fixes.
The 64tass --long-address flag doesn't appear to be necessary for
files <= 65536 bytes long, so we no longer emit it for those.
(issue #90)
Modified the asm source generators and on-screen display to show the
DP arg for BBR/BBS as hex. The instructions are otherwise treated
as relative branches, e.g. the DP arg doesn't get factored into the
cross-reference table.
ACME/cc65 put the bit number in the mnemonic, 64tass wants it to be
in the first argument, and Merlin32 wants nothing to do with any of
this because it's incompatible with the 65816.
Added an "all ops" test for W65C02.
Long operands, such as strings and bulk data, can span multiple lines.
SourceGen wraps them at 64 characters, which is fine for assembly
output but occasionally annoying on screen: if the operand column is
wide enough to show the entire value, the comment column is pushed
pretty far to the right.
This change makes the width configurable, as 32/48/64 characters,
with a pop-up in app settings.
The assemblers are all wired to 64 characters, though we could make
this configurable as well with an assembler-specific setting.
Some things have moved around a bit in app settings. The Asm Config
tab now comes last. Having it sandwiched in the middle of tabs that
altered the on-screen display didn't make much sense. The Display
Format is now explicitly for opcodes and operands, and is split into
two columns. The left column is managed by the "quick set" feature,
the right column is independent.
Added a "fake" assembler pseudo-op for DBR changes. Display entries
in line list.
Added entry to double-click handler so that you can double-click on
a PLB instruction operand to open the data bank editor.
Code generated for 64tass was incorrect for JSR/JMP to a location
outside the file bounds. A test added to 20052-branches-and-banks
revealed an issue with cc65 generation as well.
(1) Added an option to limit the number of bytes per line. This is
handy for things like bitmaps, where you might want to put (say) 3
or 8 bytes per line to reflect the structure.
(2) Added an application setting that determines whether the screen
listing shows Merlin/ACME dense hex (20edfd) or 64tass/cc65 hex bytes
($20,$ed,$fd). Made the setting part of the assembler-driven display
definitions. Updated 64tass+cc65 to use ".byte" as their dense hex
pseudo-op, and to use the updated formatter code. No changes to
regression test output.
(Changes were requested in issue #42.)
Also, added a resize gripper to the bottom-right corner of the main
window. (These seem to have generally fallen out of favor, but I
like having it there.)
Correct handling of local variables. We now correctly uniquify them
with regard to non-unique labels. Because local vars can effectively
have global scope we mostly want to treat them as global, but they're
uniquified relative to other globals very late in the process, so we
can't just throw them in the symbol table and be done. Fortunately
local variables exist in a separate namespace, so we just need to
uniquify the variables relative to the post-localization symbol table.
In other words, we take the symbol table, apply the label map, and
rename any variable that clashes.
This also fixes an older problem where we weren't masking the
leading '_' on variable labels when generating 64tass output.
The code list now makes non-unique labels obvious, but you can't tell
the difference between unique global and unique local. What's more,
the default type value in Edit Label is now adjusted to Global for
unique locals that were auto-generated. To make it a bit easier to
figure out what's what, the Info panel now has a "label type" line
that reports the type.
The 2023-non-unique-labels test had some additional tests added to
exercise conflicts with local variables. The 2019-local-variables
test output changed slightly because the de-duplicated variable
naming convention was simplified.
Implemented assembly source generation of non-unique local labels.
The new 2023-non-unique-labels test exercises various edge cases
(though we're still missing local variable interaction).
The format of uniquified labels changed slightly, so the expected
output of 2012-label-localizer needed to be updated.
This changes the "no opcode mnemonics" and "mask leading underscores"
functions into integrated parts of the label localization process.
The label localizer is now always on. The regression tests turned
it off by default, but that's no longer allowed, so the generated
output has changed for many of them. The tests themselves were not
altered.
- Renamed "strip label prefix/suffix" to "omit label prefix/suffix".
- Changed a Merlin operand workaround so it doesn't apply to code
that is explicitly not in bank zero.
- Changed {addr}/{const} annotations on project/platform symbol
equates so they line up a little better on screen and in exported
sources.
Continue development of non-unique labels. The actual labels are
still unique, because we append a uniquifier tag, which gets added
and removed behind the scenes. We're currently using the six-digit
hex file offset because this is only used for internal address
symbols.
The label editor and most of the formatters have been updated. We
can't yet assemble code that includes non-unique labels, but older
stuff hasn't been broken.
This removes the "disable label localization" property, since that's
fundamentally incompatible with what we're doing, and adds a non-
unique label prefix setting so you can put '@' or ':' in front of
your should-be-local labels.
Also, fixed a field name typo.
This adds the concept of label annotations. The primary driver of
the feature is the desire to note that sometimes you know what a
thing is, but sometimes you're just taking an educated guess.
Instead of writing "high_score_maybe", you can now write "high_score?",
which is more compact and consistent. The annotations are stripped
off when generating source code, making them similar to Notes.
I also created a "Generated" annotation for the labels that are
synthesized by the address table formatter, but don't modify the
label for them, because there's not much need to remind the user
that "T1234" was generated by algorithm.
This also lays some of the groundwork for non-unique labels.
This began with a change to support "BRK <operand>" in cc65. The
assembler only supports this for 65816 projects, so we detect that
and enable it when available.
While fiddling with some test code an assertion fired. This
revealed a minor issue in the code analyzer: when overwriting inline
data with instructions, we weren't resetting the format descriptor.
The code that exercises it, which requires two-byte BRKs and an
inline BRK handler in an extension script, has been added to test
2022-extension-scripts.
The new regression test revealed a flaw in the 64tass code
generator's character encoding scanner that caused it to hang.
Fixed.
Sometimes there's a bunch of junk in the binary that isn't used for
anything. Often it's there to make things line up at the start of
a page boundary.
This adds a ".junk" directive that tells the disassembler that it
can safely disregard the contents of a region. If the region ends
on a power-of-two boundary, an alignment value can be specified.
The assembly source generators will output an alignment directive
when possible, a .fill directive when appropriate, and a .dense
directive when all else fails. Because we're required to regenerate
the original data file, it's not always possible to avoid generating
a hex dump.
In a recent survey, three out of four cross assemblers surveyed
recommended not using opcode mnemonics to their patients who use
labels. We now remap labels like "AND" and "jmp", using the label
map that's part of the label localizer.
We skip the step for Merlin 32, which is perfectly happy to assemble
"JMP JMP JMP".
Also, fixed a bug in MaskLeadingUnderscores that could hang the
source generator thread.
Most assemblers end local label scope when a global label is
encountered. cc65 takes this one step further by ending local label
scope when constants or variables are defined. So, if we have a
variable table with a nonzero number of entries, we want to create
a fake global label at that point to end the scope.
Merlin 32 won't let you write " LDA #',' ". For some reason the
comma causes an error. IGenerator now has a "tweak operand format"
interface that lets us fix that.
The functions started by trying to pad a column out to a width,
then changed to pad things to a certain length. What they really
should be doing is padding the start of an entry to a specified
column. This is much more natural and avoids a trim operation.
The only change to the output is to ORG statements from the HTML
exporter, which are now formatted correctly.
Unlike 64tass and Merlin, which allow you to redefine symbols, ACME
uses "zones" that provide scope for local variables. This means
that, at the point of a local variable table definition, we have to
start a new zone and output the full set of active symbols, not just
the newly-defined ones. (If you set the "clear previous" flag in
the LvTable there's no difference.)
We could do a bit better by only outputting the symbols that are
actually used within the zone, similar to what we do for global
project/platform symbols, but that's a bunch of work for questionable
benefit.
Variables are now handled properly end-to-end, except for label
uniquification. So cc65 and ACME can't yet handle a file that
redefines a local variable.
This required a bunch of plumbing, but I think it came out okay.
Previously, we used the default character encoding from the project
properties to determine how strings and character constants in the
entire source file should be encoded. Now we switch between
encodings as needed. The default character encoding is no longer
relevant.
High ASCII is now an actual encoding, rather than acting like ASCII
that sometimes doesn't work. Because we can do high ASCII character
operands with "| $80", we don't output a .enc to switch from ASCII
to high ASCII unless we need to generate a string. (If we're
already in high ASCII mode, the "| $80" isn't required but won't
hurt anything.)
We now do a scan up front to see if ASCII or high ASCII is needed,
and only output the .cdefs for the encodings that are actually used.
The only gap in the matrix is high ASCII DCI strings -- the ".shift"
pseudo-op rejects text if the string doesn't start with the high
bit clear.
I didn't think it made sense, but I found something that used it,
so apparently it's a thing. This updates the operand editor to
let you choose PETSCII+DCI, and updates the assemblers to handle
it correctly (really just 64tass, since the others either don't
have a DCI directive or don't deal with PETSCII at all).
Changed the char-encoding sample from "bad dcI" to "pet dcI", and
updated the documentation.
The documentation for 64tass says you're required to pass "--ascii"
when the source file is ASCII (as opposed to PETSCII). We were
ignoring this, but it turns out that everything works a bit better
if we don't.
So we now pass "--ascii" on the command line, and add a two-line
character encoding definition to every file that is generated with
ASCII as the default encoding. The sg_petscii and sg_screen
encodings go away, as PETSCII is now the default, and we can use the
built-in "screen" encoding.
The PseudoOpNames class is increasingly being used in situations
where mutability is undesirable. This change makes instances
immutable, eliminating the Copy() method and adding a constructor
that takes a Dictionary. The serialization code now operates on a
Dictionary instead of the class properties, but the JSON encoding is
identical, so this doesn't invalidate app settings file data.
Added an equality test to PseudoOpNames. In LineListGen, don't
reset the line list if the names haven't actually changed.
Use a table lookup for C64 character conversions. I figure that
should be faster than multiple conditionals on a modern x64 system.
Fixed a 64tass generator issue where we tried to query project
properties in a call that might not have a project available
(specifically, getting FormatConfig values out of the generator for
use in the "quick set" buttons for Display Format).
Fixed a regression test harness issue where, if the assembler reported
success but didn't actually generate output, an exception would be
thrown that halted the tests.
Increased the width of text entry fields on the Pseudo-Op tab of app
settings. The previous 8-character limit wasn't wide enough to hold
ACME's "!pseudopc". Also, use TrimEnd() to remove trailing spaces
(leading spaces are still allowed).
In the last couple of months, Win10 started stalling for a fraction
of a second when executing assemblers. It doesn't do this every
time; mostly it happens if it has been a while since the assembler
was run. My guess is this has to do with changes to the built-in
malware scanner. Whatever the case, we now change the mouse pointer
to a wait cursor while updating the assembler version cache.
The 64tass generator now uses the "default text encoding" project
property to determine how readable text should be encoded. For
example, if the property is set to PETSCII, an ASCII-to-PETSCII
encoding table is generated at the top of the output file.
A delimiter definition is four strings (prefix, open, close, suffix)
that are concatenated with the character or string data to form an
operand. A delimiter set is a collection of delimiter definitions,
with separate entries for each character encoding.
This is a convenient way to configure Formatter objects, import and
export data from the app settings file, and manage the UI needed to
allow the user to customize how things look.
The full set of options didn't fit on the first app settings tab, so
there's now a separate tab just for specifying character and string
delimiters. (This might be overkill, but there are various plausible
scenarios that make use of it.)
The delimiters for on-screen display of strings can now be
configured.
The previous functions just grabbed 62 characters and slapped quotes
on the ends, but that doesn't work if we want to show strings with
embedded control characters. This change replaces the simple
formatter with the one used to generate assembly source code. This
increases the cost of refreshing the display list, so a cache will
need to be added in a future change.
Converters for C64 PETSCII and C64 Screen Code have been defined.
The results of changing the auto-scan encoding can now be viewed.
The string operand formatter was using a single delimiter, but for
the on-screen version we want open-quote and close-quote, and might
want to identify some encodings with a prefix. The formatter now
takes a class that defines the various parts. (It might be worth
replacing the delimiter patterns recently added for single-character
operands with this, so we don't have two mechanisms for very nearly
the same thing.)
While working on this change I remembered why there were two kinds
of "reverse" in the old Merlin 32 string operand generator: what you
want for assembly code is different from what you want on screen.
The ReverseMode enum has been resurrected.
The previous code output a character in single-quotes if it was
standard ASCII, double-quotes if high ASCII, or hex if it was neither
of those. If a flag was set, high ASCII would also be output as
hex.
The new system takes the character value and an encoding identifier.
The identifier selects the character converter and delimiter
pattern, and puts the two together to generate the operand.
While doing this I realized that I could trivially support high
ASCII character arguments in all assemblers by setting the delimiter
pattern to "'#' | $80".
In FormatDescriptor, I had previously renamed the "Ascii" sub-type
"LowAscii" so it wouldn't be confused, but I dislike filling the
project file with "LowAscii" when "Ascii" is more accurate and less
confusing. So I switched it back, and we now check the project
file version number when deciding what to do with an ASCII item.
The CharEncoding tests/converters were also renamed.
Moved the default delimiter patterns to the string table.
Widened the delimiter pattern input fields slightly. Added a read-
only TextBox with assorted non-typewriter quotes and things so
people have something to copy text from.
We've been treating ASCII strings and instruction/data operands as
ambiguous, resolving low vs. high when generating output for the
display or assembler. This change splits it into two separate
formats, simplifying output generation.
The UI will continue to treat low/high ASCII as as single thing,
selecting the format appropriately based on the data. There's no
reason to have two radio buttons that are never both enabled.
The data operand string functions need some additional work, but
that overlaps substantially with the upcoming PETSCII changes, so
for now all strings set by the data operand editor are low ASCII.
The file format has changed again, but since there hasn't been a
release since the previous change, I'm leaving the file format
at v2. Code has been added to resolve the ASCII mode when loading
a v1 project file.
This removes some complexity from the assembly code generators.
DCI is handled with the ".shift" pseudo-op. The .null, .ptext,
and .shift operators all work correctly with escaped characters,
so we no longer redo those.
This generalizes the string pseudo-operand formatter, moving it into
the Asm65 library. The assembly source generators have been updated
to use it. This makes the individual generators simpler, and by
virtue of avoiding "test runs" should make them slightly faster.
This also introduces byte-to-character converters, though we're
currently still only supporting low/high ASCII.
Regression test output is unchanged.
We used to use type="String", with the sub-type indicating whether
the string was null-terminated, prefixed with a length, or whatever.
This didn't leave much room for specifying a character encoding,
which is orthogonal to the sub-type.
What we actually want is to have the type specify the string type,
and then have the sub-type determine the character encoding. These
sub-types can also be used with the Numeric type to specify the
encoding of character operands.
This change updates the enum definitions and the various bits of
code that use them, but does not add any code for working with
non-ASCII character encodings.
The project file version number was incremented to 2, since the new
FormatDescriptor serialization is mildly incompatible with the old.
(Won't explode, but it'll post a complaint and ignore the stuff
it doesn't recognize.)
While I was at it, I finished removing DciReverse. It's still part
of the 2005-string-types regression test, which currently fails
because the generated source doesn't match.
The 65816 definition makes it a two-byte instruction, like COP. On
the 6502 it acted like a two-byte instruction, but in practice very
few assemblers treat it that way. Very few humans, for that matter.
So it's now treated as a single byte instruction, with the following
byte encoded as a data value.
This worked, sort of. The problem is that SourceGen will revert to
hex output in certain situations, such as a broken symbolic
reference. There happens to be one in the ZIPPY example, and it's
on a relative branch.
The goal with the segment stuff is to allow cc65 to treat the
source as relocatable code. In that context, a relative branch to
an absolute address doesn't make any sense, so the assembler reports
a range error.
We don't currently have a mechanism that guarantees no references
are broken (and no affordance for finding them), so we can't make
this mode the default yet.
Instead, we continue to use the generic config, but generate the
correct set of lines as comments.
(issue #39)
