The old code included the @ in mention color computations. If the nick
consisted only of weird unicode characters, this resulted in an
incorrect color being computed.
When the bound matches the implicit bound, i.e. when you'd just write
impl ... + use<'_>
then it can be omitted. My gut instinct is to always write the bound
explicitly, but maybe that'll harm readability once I'm more used to how
bounds work now.
Anyways, always try to keep the bound as small as possible, ideally just
impl ... + use<>
This change brings them in-line with the default settings of
rust-analyzer. Originally, I disliked this style, but by now, I've grown
used to it and prefer it slightly. Also, I like staying on the default
path since I usually don't need to check the imports to see if
rust-analyzer messed anything up (except for omitting the self:: when
importing modules defined in the current module, grr :D).
I've also come around to the idea of trailing mod definitions instead of
putting them at the top: I would also put module definitions that
contain code instead of referencing another file below the imports. It
feels weird to have code above imports. So it just makes sense to do the
same for all types of mod definitions.
If rustfmt ever gains stable support for grouping imports, I'll probably
use that instead.
Except rusqlite and vault, because newer sqlite versions appear to
result in a *very* big performance drop. I suspect the query planner,
though I really have no idea.
