Stories of source code loss
Added 2022-04-26 13:01:02 +0000 UTCIt's funny how often source code gets lost.
Once upon a time, I went on a few dates with a writer who worked at Bioware Austin. Of course, like almost every game studio, it was a famously bad place to work that also had terrible development practices.
Something that surprised me at the time was that they'd lost the source material for older games, so writers would play through older games when working on a sequel to be able to read the text. I think that was the first time I heard about source code/material loss, so it was shocking to me. And it's also odd that a programmer didn't rip the text from a binary so that writers could get easily read the text.
Since then, I've heard enough stories about source code loss that I've stopped being surprised by them. Even places that aren't really famous for having sloppy practices also sometimes have sloppy practices that lead to source code loss.
For example, someone who worked on Larrabee told me that the project was much more difficult than it had to be because Intel lost the Verilog source code to the P54C core that Intel used for Larrabee. Luckily for Intel, they'd given a version of the source code converted to VHDL to a partner. That version had all sorts of stuff wrong with it (truncated names, incorrectly mangled names, etc.), but it had the advantage of actually existing, so they converted the VHDL back to Verilog and, after fixing it up, used that as the source code for Larrabee cores.
While Intel is known for having inefficient practices in a variety of ways, being sloppy enough to lose the source code to one of their old designs doesn't really fit their reputation the same way it fits Bioware's reputation, but Intel managed to do it. And likewise for at least one useful chunk of code for at least one major tech company I've worked for.
