On Monday, 6 October 2025 at 13:54:12 UTC, jmh530 wrote:
I'm curious if you are aware of Retrieval-augmented generation
(RAG) [1], or know if any of the other major D developers are
aware of it.
I'm not aware of that specific term but I am aware AIs can be
specialized for specific domains. I haven't done anything like
that myself yet.
So for instance, suppose we create a custom D GPT, then feed in
the standard for 80bit floating point (maybe also an example of
another type of emulator), then ask the custom GPT to write an
emulator for it. See what happens.
80 bit floating point is almost completely decoupled from dmd and
could be implemented in C, I don't see why the AI needs
D-specific knowledge to write such an emulator. But the idea is
interesting, and indeed maybe it can help improve the backend's
debug info emission.
Even without a custom GPT it's already nice that I can ask an LLM
how DWARF works, and on https://deepwiki.com/dlang/dmd I can ask
where the DWARF emission is currently implemented in dmd. I
remember at DConf '22 asking Luís Ferreira about enum member
debug info generation because online I couldn't easily find info
on how to do that.
But still, I'd be interested in seeing AI specialized in dmd / D,
it's just not something I myself plan to look into short term.
