xflags VS config

3) xflags is my preferred way to parse CLI arguments for such scripts. It's not good at all, but, for me, it's the best for the task at the moment.

You might like xflags. It has a similar design philosophy, but is willing to compromise on simplicity a bit to get auto-generated help.

See also https://github.com/matklad/xflags/issues/3

Added https://github.com/matklad/xflags/blob/master/examples/non-utf8.rs.

I actually quite like the config, seems intuitive! Though, I am comparing it to this abomination, so I might have low expectations here :-)

4) For scripts you put for everyday use in ~/bin, I've found busybox style multipurpose binary invaluable. You put all your scripts into a single Rust binary, and than hard-link this binary under different names in PATH, so that the first argv is actually the name of subcommand. Here's how the infra works for my scripts, and here's an example script.

1) Invest into application-level profiling, using tracing::span! for collection and something like tracing_tracy or tracing_span_tree for visualization. This should give you high-level logical profile. 2) If on Linux, learn a bare minimum of perf for collecting data (the extent of my knowledge about perf is basically this script https://github.com/matklad/config/blob/master/tool/src/prf.rs). Use neither flamegraph.pl nor cargo flamegraph for visualization, use https://profiler.firefox.com/ instead.

And, while a spam links, here’s a link with more links: https://github.com/matklad/config/blob/master/links.adoc