@emily_s AHHH OMG THIS IS SO COOL
Discussion
Loading...
@emily_s i have a parallel directory traversal crate which is faster and more configurable than the ripgrep ignore crate (which uses walkdir for single threaded iteration) but it's not done yet
@hipsterelectron @hipsterelectron Thanks!
And thanks for the idea of having the config functions make a tree. That made this much more fun :D
If I get to the point that I'm worried about the performance of walking the directory structure I'll call that a champagne problem as clearly this has got a bit out of hand
@emily_s i will tell you i have spent months on the directory traversal crate in order to achieve meaningful perf improvement on massive inputs. useful for very fast parallel zip archive creation and potentially almost nothing else
@hipsterelectron also probably useful for making one of those file system visualisations that show you where all your disk space has gone.
@emily_s yes!! indexing a whole file system like locate/updatedb was absolutely one goal i considered although making a find clone is also interesting. the findutils project has wonderful documentation and is very nice to me over email
@emily_s i'm also p sure the tree structure was your own insight
@emily_s sometimes doing stuff like the parallel traversal crate is partially motivated by achieving a clear negative result because it can be extremely useful to be able to say that i (or someone) have deeply investigated this possibility and it's not worth spending time on
@hipsterelectron I can understand that. That's partly why I started down this code config path. Why are we templating yaml files when we have a perfectly good programming language over there? There must be a good reason
1+ more replies (not shown)
@emily_s walkdir's clear semantics are what you want in this case though and its impl is rock solid