Thread by @a2_4am

Story time.

In 1981, @jmechner tried and failed to publish an Asteroids clone for the Apple II. It looked like this:

38 years later (so, this year), @a2_qkumba ported it to ProDOS and we included it in "Total Replay," our pack of 200+ Apple II games <archive.org/details/TotalR…>. Our version looked like this.

Shortly thereafter, someone reported a bug against Total Replay that Asteroid Blaster had corrupted graphics. This is ironic since Jordan had gone to great lengths to ensure that the asteroids looked exactly like the original, and this is why the game was ultimately canceled.

Plot twist! After minutes of furious research, @a2_qkumba realized that this graphics corruption bug was NOT introduced in 2019; it was present in the original version from 1981. Due to an off-by-1 bug in a branch instruction, the game was landing on an undocumented opcode.

On the original 6502 CPU (Apple ][ and ][+, and early //e models), this bug was harmless; the undocumented opcode did nothing, and the bad branch went undiscovered.

But on a 65c02 CPU ("enhanced" //e), this opcode got redefined and ultimately led to visible graphics corruption.

Obviously @a2_qkumba fixed it, applying a binary patch to Jordan's 38-year-old—and apparently hand-coded—6502 code to fix the branch, avoid the undocumented opcode, and fix the graphics corruption when running the game on a 65c02.

The fixed version will be in Total Replay v3.

qkumba also fixed the original self-booting disk image, which I have uploaded to the Asimov FTP server as "Mechner_Collection_Astroid_Blaster_and_Star_Blaster (65c02 compatible).dsk"