Spectrum game disassembly toolkit

Code > text > data

SkoolKit 7.1 has been released. As usual, copies may be obtained from the download page, the Python Package Index, the Ubuntu PPA, or the Fedora copr repo.

Among the changes worthy of note in this release is the improved control file generation algorithm that uses when no code map is provided. The algorithm is now much faster, and strongly prefers to mark a chunk of bytes as code first, text second (if it really doesn’t look like code), and data last (if it really looks like neither code nor text). The main idea behind these preferences is that more code and text will be identified, at the expense of misidentifying more data. But since data that has been marked as code or text is easier to spot and fix than code or text that has been marked as data, the cost is worth it.

In other news, that command can now read configuration from skoolkit.ini, and has also gained the --ini option (for setting the value of a configuration parameter) and the --show-config option (for showing all configuration parameter values). So if you want to always produce hexadecimal control files, you can add the line Hex=2 to the [sna2ctl] section in skoolkit.ini and forget about having to use the --hex option on the command line all the time.

Speaking of control files, now has the ability to read multiple control files by default (e.g. both game-1.ctl and game-2.ctl when the input snapshot is game.z80), and to use the --ctl option multiple times (for specifying two or more control files manually). This change makes easier to use if you prefer to split your large control file into smaller, more manageable chunks.

On the image macro front, #UDGARRAY can now specify attribute address ranges, as an alternative to specifying an attribute value separately for each tile. For example, if you are creating an 8x4 graphic whose attribute values are located at addresses 32000-32031, you can now write a macro like this:


Hopefully we can all agree that this is much nicer than having to specify each of those 32 attribute values individually.

And that’s it for the new features that I can be bothered to spend time explaining in any detail. If you want to know more about the other changes in 7.1, such as the ability to render numeric values in instruction operands as negative numbers (e.g. LD A,-1), or how the @*sub and @*fix directives can insert an instruction after the current one without first specifying a replacement, I suggest heading over to the changelog.

Surviving the round trip

ctl-invariance ftw

SkoolKit 7.0 has been released. I invite you to visit the download page, the Python Package Index, the Ubuntu PPA, or the Fedora copr repo if you would like to collect a copy.

As you might expect with the initial release in a new series, there are some changes that break compatibility with previous versions. The two main ones are that the ability to generate a control file has moved from to the new command, and that now expects a skool file as its first positional argument, and treats any remaining positional arguments as extra ref files. For details on the other ways that SkoolKit 7 might break your disassembly, and how to fix it, see the migration guide.

But enough of that. What new features does this release bring? Well, again, as with the compatibility-breaking changes, there are two main ones, and they both serve a purpose that has long been dear to my heart: preservability in a control file. It has always irked me that some parts of some skool files are not amenable to such preservation, but with the advent of 7.0 those parts (and those skool files) are thankfully smaller in number.

First of all, the @isub, @ssub, @rsub, @ofix, @bfix and @rfix directives have gained some key new abilities. They can now specify a replacement comment over multiple lines, replace labels, and insert, overwrite and remove instructions. Previously, each of these operations would have required the use of a block directive (e.g. @ssub-begin@ssub-end), which cannot be preserved in a control file.

Second of all, non-entry blocks (i.e. blocks that do not match the format of an entry, such as a header comment) are now preserved verbatim by, and are reproduced verbatim (minus any ASM directives) by This means that comment blocks appearing between entries in a skool file will now survive a round trip through and, even if they contain ASM block directives.

Third of all, although I said there were only two main new features, there is one more that is relevant to the central theme of preservability in a control file, and so deserves a mention here. Inverted characters - meaning character codes with bit 7 set to indicate the end of a string - can now be stored in and restored from a control file. So, for example, the "o"+$80 in DEFM "Hell","o"+$80 will survive the round trip, instead of degrading to $EF.

And that is it, as far as preservability in a control file is concerned. For details on all the other much less important features - which, I warn you, may have nothing to do with either preservability or control files - take a look at the changelog. Happy migrating!

An @if but no @but

Do not expand this macro

SkoolKit 6.4 has been released. To get your copy, pop along to the download page, the Python Package Index, the Ubuntu PPA, or the Fedora copr repo.

Continuing the trend of new ASM directives that was started in 6.3, this release brings us the @if directive. As its name implies, it can be used to conditionally process other ASM directives. The variables it can use are the same as those available to the #IF and #MAP macros: asm, base, case, fix (new in 6.4), html and vars (also new in 6.4). So, for example, if you want to replace ‘foo’ with ‘bar’ in HTML mode, but with ‘baz’ otherwise, you could do this:


As noted above, the fix and vars replacement fields are new in this release. fix indicates the fix mode (0, 1, 2, 3 for none, @ofix, @bfix, @rfix), and vars is a dictionary that holds values defined by the new --var option of and In addition, asm now indicates the exact ASM substitution mode (0, 1, 2, 3 for none, @isub, @ssub, @rsub) instead of just being 0 or 1.

Still on the subject of ASM directives, @isub, @ssub, @rsub, @ofix, @bfix and @rfix can now replace not only the instruction on a given line, but also the comment that goes with it. This capability reduces the need for using ASM block directives (@bfix+begin etc.), which cannot be preserved in a control file.

Moving on to the skool macro department, this release introduces the second and final new macro of the 6.x series: #RAW. If you’ve ever been annoyed by SkoolKit’s inability to render a hexadecimal number higher than 40959 (0x9FFF) in upper case with a ‘#’ prefix (e.g. #ABCD), #RAW will quickly become your new best friend. It prevents any macros or macro-like tokens in its sole string argument from being expanded. So you can now write #RAW(#BEEF), safe in the knowledge that SkoolKit will not complain about ‘#BEEF’ being an unknown macro.

Finally, a word about the venerable #LIST and #TABLE macros. In previous versions, neither of these macros could be used in an instruction-level comment or as a parameter of another macro in ASM mode. That restriction has been lifted in 6.4. Since no one’s ever complained to me about it before, I must conclude that either no one noticed, or no one really cared. But preventing the use of these macros in those locations in that mode made no sense, so there we go.

For details on the other new features in this release, I will point you to the changelog. I would also recommend taking note of all the deprecated features, so that you’ll be ready for 7.0 when it arrives. See you then!