The "Ada" topic is (only) for the last part. It precisely because with such a good tool, you focus on the important things: the algorithms.
Precisely, there is no such a thing as the BZip2 algorithm: there are multiple ways of programming it, and more importantly, there is a near infinity of possible encodings of the same inputs in the BZip2 format. The compression ratio depends on the choices made in that area. That point is addressed in part 3.
Unfortunately the author of the TIOBE index has obeyed too quickly to what Mr Herrera told him. The hits about that "Ada Lovelace architecture" are a negligible part of the hits you get with: +"Ada programming" +"nvidia". The lions' share is about Nvidia's own usage of the Ada language. Those hits should be counted of course.
Thanks for the feedback. Multiple posts are planned on that topic. For parts 1 and 2 (so far the only ones existing) I preferred explaining the context and the remarkable simplicity of the BZip2 format. In part #4, I will try to show why the customisable types in Ada fit so well compression software.
Be patient: writing articles about writing software takes much more time than writing the software itself!
In the meantime, there is a short preview of that in recent presentations - notably FOSDEM 2025: https://fosdem.org/2025/schedule/event/fosdem-2025-5148-adve... , last slide.
Interesting letter!
His concerns were legitimate regarding complexity, but most of the items in his list of features are indispensable.
Basically Ada 83 was a pre-version lacking key features for practical use, although it had already cool features (exceptions, modularity, generics, tasks, ...).
* Splitting large enumerated types into smaller ones
* Using the same range for arrays indexes, "for" loops parameters which index those arrays. In those cases a good compiler removes the useless run-time checks!
Ada range types can have bounds that are known only at run-time. It was not possible with Pascal.
To see subtypes in action: another FOSDEM presentation:
