Gosh, I love #Libretto’s so much but they’re basically unobtainium at this point and becoming ever more fragile. https://youtu.be/xfMrqJwuMX8
I like trying to find the balance between minimal and usable in old tech. I have a beautiful Dell X200 which just straddles the line of obsolete and modern as even though it’s an 800MHz PIII it can connect to WiFi and run a browser barely new enough to access the web. #RetroComputing
Even though old software has not changed, new software has introduced faster, more efficient workflows at the cost of the speed of the software itself. The knowledge and tools to even produce software that uses a few megabytes of RAM is, in practice, lost. We can produce better tools but we cannot produce them for the computers that would benefit most from them. #RetroComputing
Have you ever tried to compile an old piece of software that first required compiling an old compiler which itself required an old environment that you can't reproduce? #GCC & #LLVM all work within very narrow bands of time, platforms and environments that shift constantly, meaning that the aparant portability of C/C++ is unusable in practice for any non-current OS. #retrocomputing#cpp
I don’t think developers truly appreciate the fragility of software until their first run-in with an obscure file format for which the only converter in existence was written by one guy in college in 2003 and released as source only with no comments and requires applying a patch to a library to compile, but it’s not for a release version, oh no, it’s for a very specific SVN revision from a long dead server
Ah, but I’m speaking of the Before Times; good luck finding that one student’s public FTP folder using current Google, and I can assure you ChatGPT doesn’t know the answer. Internet Archive has the page, but not the zip file.
Software fragility takes many mundane forms. Have you tried running a python script that uses an unknown version of Tensorflow from a mere 5 years ago? We ended up rewriting it from scratch.
Add comment