[Nix-dev] Re: Proposal: tryCatch function

[Andres Loeh]

> We have enabled side effects already. Without caring about evaluation
> order. I mean "trace" builtin. It has 2 main properties, though: it
> never changes what derivations are constructed and it only gives
> end-user a bit more information that can be filtered out if not helpful.
>
> tryCatch lacks the first property, and because of that I consider it
> harmful. tryRethrow is just like trace, and I am not sure that IO purity
> is worth preserving (if it doesn't affect actual build purity).

As you say, trace is a somewhat special case. Nix is a special-purpose
language, and the results we're interested in are the derivations, not
the screen output. So having some slight impurity that affects only
screen output and nothing else is acceptable IMHO.

Regarding tryCatch: as far as I understood it, it doesn't introduce any
impurity, so I don't find it particularly problematic to add. But I doubt
its usefulness given the lazy nature of the language and the escaping
issues. Plus, I think it sends the wrong message, because it
encourages the use of
exceptions (throw), which are impure already, and I don't like that very
much. I think we should strive for ways to be able to program with as few
as possible uses of throw. On the other hand, I'm willing to admit that
while we're using throw anyway, it might be better to at least provide
helpful error messages.

Cheers,
  Andres