Conversation

Something I’ve recently realized: a running program is just a static program with some extra data (an event history). There’s no reason we shouldn’t be able to fork, edit, undo & redo running programs (and archive them) just as we do with static ones.

116

Brendan Zabarauskas

@brendanzab

Aug 26

Oooh yeah! Sounds similar to what declarative debugging is based on! I think there were papers on this on implementations for Haskell and Mercury.

Brendan Zabarauskas

@brendanzab

Aug 26

Also related might be the interaction trees stuff for Coq (more from a proof perspective).

Replying to

and

I think there the use case there iirc is proving stuff forall possible ‘traces’ of a effectful programs?

11:48 AM · Aug 26, 2022Twitter for iPhone

Replying to

The experience I proposed can only be built upon a declarative language, so there's a connection there. But I don't otherwise see a strong link to the research you've mentioned—it seems to be about a *better* debugging tool, whereas I'm saying we should mash everything together!

Brendan Zabarauskas

@brendanzab

Aug 26

Oh yeah, that's a specific use case definitely and far more limited compared to what you are proposing. More thinking these are different things based on this way of looking at programs… if that makes sense. Sorry!

Show replies