Differences between versions

K240 was released in two main versions: v1.886, the original retail release with a build date of 20 May 1994; and v2.000, which appears to be a bug fix, build date 7 June 1994. You can press “V” during play to find out which version you have.

Additional non-retail versions of K240 are:

• A demo version distributed with CU Amiga issue #77 (Mar 1994) as an exclusive coverdisk. It appears to be based on an earlier build of the game circa January 1994 and has some differences: some notable bugs, and different graphics. I intend to disassemble this for comparison eventually.
• A demo version available by limited mail order offer in Amiga Dream #7, which may or may not be identical to the CU Amiga demo.
• At least one elusive version distributed to Amiga magazines as a review copy, mostly finished, and probably with a build date March 1994.
• Various cracked pirate releases of v1.886 and v2.000. Aside from removed copy-protection, cracktro, and trainers, these seem to be largely unchanged.

1. Main game executable differences
   1. Fleet bug fix
   2. Version strings
   3. Padding
2. Cracked version vs original
3. Directory comparison between v1.886 and v2.000
4. Directory comparison of the CU Amiga demo
5. Executable analysis of the CU Amiga demo
6. Other demo versions
   1. Amiga Down Under #9 Coverdisk demo
   2. Amiga Dream mail-order demo
7. Physical media differences
8. Elusive Amiga magazine review copies

Main game executable differences

The following differences apply between the v1.886 and v2.000 versions of the main playk240 executable file.

Fleet bug fix

The main game executable playk240 appears to have only one functional change between the v1.886 and v2.000 release, which is a single additional line of code to address a major bug that crashes the game when handling fleets.

At address 0xc0d8 in v2.000 (0xc0da in v1.886), at the end of a loop, it checks the value in (6,A0), leaving the loop on a negative value; otherwise, it loads a new value into (6,A0) and repeats the loop. In v2.000, it an extra BNE.W instruction is inserted at 0xc0de between the BMI.S and MOVE.L, which repeats the loop without updating (6,A0) if it is equal to a non-zero value. This extra four-byte instruction increases all addresses after 0xc0de in the executable by 2.

Version strings

The version string at 0x6016 in both versions is also updated. This can be seen in-game by pressing the V key. The two version strings are:

K240 - VERSION 1.886,  20-5-94 13:25
CLICK MOUSE TO CONTINUE GAME...

K240 - VERSION 2.000,  7-6-94 11:15
CLICK MOUSE TO CONTINUE GAME...

The v2.000 version string is two bytes shorter overall.

Padding

At the very end of the first hunk, after the cheats, the v2.000 disassembly has an additional two bytes, 0xb900. This is never called by the program and is probably just padding to align the next section.

Overall, v2.000 gains four bytes from the fleet fix and two from the padding, but loses two from the shorter version string. The result is that v2.000 is four bytes larger overall.

Cracked version vs original

K240 has a manual protection code system which requires the entry of one of a hundred words from the game’s manual. The authorized free download of K240 is based on a cracked pirate version of the v1.886 release, which made a single two-byte change allowing any manual protection code to be accepted.

The change occurs at location 0x40ee (decimal 16622 bytes) in the playk240 executable, or at 0x40c6 in the disassembly. It is the same location for both v1.886 and v2.000.

The original instruction is MOVE.W #$025b,D0 (hexadecimal 303c 025b) , which places the ID number of the text string “AUTHORISATION CODE INVALID!!” into data register D0. The cracked version has overwritten the first two bytes, 303c, with 6016, which branches forward 24 bytes to the instructions normally executed on entry of a correct code. All incorrect codes are therefore effectively treated as correct codes.

The authorisation code requirement can be enabled or disabled by hex editing the two bytes at location 0x40ee in the playk240 executable. Setting the value to 6016 disables it, which may be useful if you have a retail version but have lost your manual. This is especially useful if you own a v2.000 copy of K240, since the official download is v.1886 and suffers from the fleet bug.

Changing the two bytes back from 6016 to 303c will restore the executable to its original state. You can verify this by running the checksum program on K240 disk 3, which will give a value of $1cc3 for an unmodified v1.886 playk240 file or $2471 on v2.000, although the program may fail on a cracked version due to other modified files, such as a packed intro file on disk 1. I’m not sure why you would actually need to re-enable the manual code system, but perhaps you need an unmodified original executable for speedrunning, or just want to see what it looks like when you input an incorrect code.

It is interesting to note that string 0x25b, “AUTHORISATION CODE INVALID!!”, is one of the last entries in the text string list. Only two strings appear later in the list, which are 0x25c, “AUTHORISATION CODE ACCEPTED…”, and 0x25d, “ENTER A NAME FOR THE SAVED FILE:”. This suggests that the copy protection system was added very late in development.

Directory comparison between v1.886 and v2.000

My version history for Dungeons of Avalon went more thoroughly into the file differences between various releases. However, K240 versioning is relatively simple. It did not write save games to its own disk, and in fact had a feature to prevent this, so most K240 disk images are unmodified. There are fundamentally only two versions: v1.886, and v2.000.

Here I’m comparing the v1.886 authorized download from Gremlin Graphics World to the v2.00 SPS .ipf release. The authorized release is cracked by TRSI & Zenith (TRZ); I assume the IPF to be unmodified from retail.

The file date on the Gremlinworld ADFs is 3 Sept 1997, suggesting someone ripped these ADFs very early. CU Amiga magazine still existed when someone made ADFs of their pirate copy of K240. The UAE emulator was only about two years old. The Web Archive suggests that this zip had already been uploaded to the Gremlin Graphics World by 14 August 2001.

It’s clear that only file which changed between v1.886 and v2.000 is the main game executable, playk240 on disk 2. In other words, none of the supporting game files actually changed. A convenient checksum program for this very purposes exists on disk 3. Now, the checksum program on disk 3 also changed in v2.000, but only to update the checksum for the newer build of playk240.

There are some other differences, based on the file dates and md5sum. These are limited to disk 1, and appear to all be caused by the pirate copy inserting its cracktro message.

The cracktro program TRZ was added, and the startup-sequence modified to add it. The intro executable shrunk from 20,740 bytes to 10,524, which appears to be TRZ compressing it to make room for their 2,920 byte cracktro. Disassembly shows that the smaller version is obfuscated while the larger isn’t, suggesting that it is indeed compressed. The v1.886 checksum program does not recognise these changes, so we can infer that the compression is not part of the original disk.

The file dates on these give us a good sense of just how rapidly the game was pirated. They’re all dated 26 May 1994, while the build date embedded in the executable is 20 May 1994, a Friday. This means the game was cracked and pirated within six days of being finished, or four working days, including printing and shipping the disks. The cracktro credits the supplier of the original disk to “Troops & Troll”.

My original hypothesis had been that v1.886 was an early build for magazine reviewers, as there is a rumour that a staff member at one of the Amiga magazines often leaked review copies to pirate groups, leading some games to be pirated even before street date. However, on further analysis I believe it to be the retail version, since magazines with reviews would have already been published before the v1.886 build date (see development and release history). The review build likely had an earlier version number.

Many files are dated 01-Feb-78, which looks like a default Amiga file date for computers with no real-time clock. Notably, however, not all files are. Various files are dated 20 Sept 1991, 2 Sept 1992, 2 Mar 1994, and 16 May 1994. It’s possible that many of the game’s graphical assets were already in place in September 1991; however, I suspect that these are default dates from an Amiga with a hard disk but no real-time clock, which would therefore take its time from the hard disk’s OS install date on boot-up.

Directory comparison of the CU Amiga demo

All file dates on the CU Amiga coverdisk #77 demo of K240 are given the false default date of 1978, so all we can say for certain is that it shipped with the magazine dated March 1994, probably released in February 1994 as was the custom, meaning that the disk was already completed in January or February 1994, at least three months before the v1.886 retail build.

The 1978 dates actually span several months, so we might infer that whoever wrote these files used an Amiga which was left switched on for several months at a time. The latest is dated 20 July 1978.

Obviously, not all of the files from the full game are present on this one disk.

All six alien and planet graphics are there, unchanged, as are the six shop blueprint files and several backgrounds: wireplan.mgl, hologram.mgl, and tetra.mgl.

However, there are no voice files, outro graphics, or language files. welcome.mgl, which I believe is the “welcome to sci-tek” audio, is not there, and nor is satpic.mgl, which isn’t needed since the demo’s spy satellite feature is disabled. None of the alien data files are there (you can only fight one alien in the demo, although the English language strings for the alien descriptions are still in the game code).

There are two unique demo files: demo.mgl and demo2.mgl. These are probably the unique intro and outro screens. The file scitek.mgl is present, but it has changed; it’s 18,052 bytes instead of 18,035, for reasons I have yet to discover.

Executable analysis of the CU Amiga demo

The demo version of K240 distributed exclusively with CU Amiga magazine is 451,998 bytes, which is 17,780 or 17,776 bytes smaller than the final releases. However, an in-depth analysis of the code needs to be done to determine the differences between this demo and the final version.

Other demo versions

Amiga Down Under #9 Coverdisk demo

Amiga Down Under issue #9 (May 1994) included a coverdisk with a demo of K240. It is functionally identical to the CU Amiga demo, even including the special graphics declaring the game a CU Amiga exclusive. The disk has a creation date and time of 06-Apr-94 12:38:42, which is after CU Amiga released their coverdisk on the March 1994 issue. The CU Amiga disk has a false creation date of 1978 caused by being created on a standard Amiga without a real-time clock. There are a few minor quirks about the exact disk structure, but effectively all files and all file dates on the Amiga Down Under coverdisk are exactly the same as the CU Amiga dates.

Amiga Dream mail-order demo

Amiga Dream issue 7 (May 1994) not only reviewed K240, devoted the following page to a mail-order offer of 300 playable demos of K240 to give away. It’s unknown what exactly this demo contains. There’s a chance it is just an identical copy of the CU Amiga demo. However, it could plausibly be a unique build with French language.

If you have a copy, it would be a great help if you can respond to this ticket in the project’s issue tracker.

Physical media differences

Some K240 disks have the label printed in orange, while others are printed in white. Based on photos of copies of the game on eBay and elsewhere, the “white” disks were accompanied with the K240 tutorial VHS flyer and a Gremlin prize draw entry card, and their boxes had stickers attached advertising high review scores (CU Amiga 91%, The One 90%). Both versions were widely available.

The most plausible explanation is that the white-label disks are from a second print run. I have a hypothesis that all orange disks are v1.886, while white disks are v2.000.

The Polish release of the game was published by a separate company, Licomp, and had its own unique box art as a result, which is largely a recreation of the English box art. Likewise with the manual. The game itself is probably identical to the English build, and not a Polish translation. Polish magazine Świat Gier Komputerowych, Jul/Aug 1994, which featured an original render of K240 for its cover, shows only English in screenshots, and the game came with a list of the English manual protection codes.

Elusive Amiga magazine review copies

At least one build of the game was sent to Amiga magazines as a review copy in March 1994, so that the magazines could review the game in time for their May 1994 issue. It’s not the same build as the retail release, because that wasn’t completed until 20 May 1994, and magazines released their May issues in April.

Only a tiny number of review copies were distributed, and the fate of the disks after the magaziness closed down is unknown—if you have a copy, please respond to the ticket on the project issue tracker, as you’d be a major help to the project.

What we do know about the review build is as follows:

• It is nearly complete to the retail v1.886 version. Screenshots show the final Construction Yard and Environment Control sprites, rather than the early sprites in the CU Amiga demo.
• The hard disk installer is included. This suggests a date of no earlier than 2 March 1994, which is when the HD installer was added in the final release. The installer adds files by folder and wildcard, so the addition of files later in the game would not have necessitated changes to the installer. Given magazine deadlines, the review build would have been made in March 1994, about two months before the finished v1.886 release.
• It is English-only. A review in French magazine Joystick issue 49 (May 1994) shows only English-language screenshots (as do other French magazines’ May 1994 reviews of K240), but promises that the final release will have a French translation.
• The victory screen background uses the intel screen background as a placeholder, as seen in The One issue 68. Since MGL image file sizes have to be defined in the executable, this means the review build executable has to be different to the final build.
• The intro disk appears to be included. Voice lines are in the game, according to the Joystick review. This suggests the review copy included all three disks.
• The finished manual appears to have been distributed with the review copy.
• Magazines which reviewed K240 for their May 1994 issue are assumed to have received an advance review copy. These included the English-language magazines Amiga Action, Amiga Computing, Amiga Format, Amiga Power, CU Amiga, and The One Amiga; and the French-language magazines Amiga Concept, Amiga Dream, and Joystick. If you have any disks which survived the closure of those magazines, please let us know.