Tim Cuthbertson

> It would be nice if the user could just add a direnv cache directive into his .envrc to enable caching That would be nice, but yeah: > Change tracking...

> The only downside is that direnv's execution time on every prompt is linear to the number of files. It's probably fine on SSDs but what if for example the...

Ah, thanks for figuring that out. Agree it's confusing, if we're going to change them perhaps the flag which sets the generated name could be `--output-entrypoint=...` (to match `--output`), and...

Updated to use magenta @chtenb . I think grey/default might be confused with other unfiltered output, it's good to have it be coloured in some way.

Oh and files in the current directory (or just outside the stdlib? not sure) are already printed in grey, so it'd good to maintain a different colour for the prefix.

It's green now (https://github.com/koka-lang/koka/commit/0c7d5133eac8447af11fbd5f025c4fdd9c49b6fa) (which I like fwiw): 

One related feeling I have (after only modest use, mind you) is that div is quite prominent in code, yet it's rarely something I think about. I absolutely care if...

This is a bit out there and probably not practical, but could there be opt-out effects as well as opt-in? If `div` were opt-out, then everything would be div unless...

Thanks for raising this, the readability issue was a surprise to me. My original line of thinking was "why would you set this to an unreadable path"? But of course...

> Another solution would be to have the process override NIX_SSL_CERT_FILE to point to the location where the contents of caFile were placed inside the chroot. This is better. Yes,...