On Mon, Feb 09, 2026 at 10:07:11AM +0100, Fabian Groffen via Mutt-dev wrote: > On 09-02-2026 00:51:00 +0100, Gero Treuner wrote: > [...] > > I generalized the patch to cover subpaths of different lengths and also > > applied the optimization to mx_is_mh() . > > > > IMO code clarity is sufficient. The array declarations could be moved > > inside the function, if that might further improve it. Repeated function > > calls are avoided, CPU cycles probably similar, and size of binary > > slightly reduced: ~100 Bytes for stripped mh.o and also affecting the > > final binary (not stripped) on my Debian Linux amd64 machine. But I > > assume that with different compilers, optimization flags and machines > > results vary. > > I decided against the proposed fstatat(), because most likely there are > > some mutt users on machines not supporting it. Adding detection and > > alternative code should be overengineering. > > > > The code is custom-made to fit to the current buffer code. (Extending it > > would not be an option in an exercise to minimize code.) > > I like this a lot. It's clear and easily extendable.
Me not. And as Kevin already committed his version to the master branch, I don't see much value in discussing this any further. But this is just my 2 cent.
