2026-08-17 Test driving the FreeBSD fixed niri wm
I had mentioned earlier about a possible set of fixes for FreeBSD and got some time to sit down and test drive the repo. I tested this on a 15.0 FreeBSD (will test again also after getting to 15.1) that was available. The available documentation inside of the freebsd/ folder was a nice touch. However, it seems that the author of the patches made an effort to use as much available in upstream as possible but unfortunately I ran into problems...
One of the first issues is inside of the apply.sh script that was made available. The troublesome lines involve the logic around applying the patches that were also included. It seems that the author must have installed GNU patch at some point and assumed that a typical FreeBSD user would also have it installed. While I will keep it short, in general assuming GNU cli utilities is usually a sign of inexperience at working with non-Linux environments which in this day and age is actually probably more common than expecting the multi-UNIX distro (as in Solaris, HP/UX, BSD, and Linux) fluency that is more of a sign of being an old fogey in 2026.
I have lost count of the number of times I crawl through bash-isms in something claiming to be /bin/sh then on top of which the bash script decides to use a particular ls command line option that only exists on some mumble mumble version of G** utilities. I guess I have gotten numb to this unawareness.
After fixing the patch, it looks like I then ran into having to double check the patching of the rust crates which did finally apply. Then realizing the nice little start script has nvidia hard-coded which doesn't fare well on something like a Framework AMD board so applied a little Go-based approach (a little copy here or there...) to 'fixing' it finally got it to run.
Based on all the changes made in the repo on top of my own little tweaks, I can only hope these get merged into upstream but it looks like to apply the fix requires updating a small handful of library crates, possible tweaks to FreeBSD itself, along with the niri wm too feels like quite a bit to ask for what I originally (mistakenly) assumed was a small patch to just the wm itself.
I have published my changes here if people just want to pull down those changes. My hope is the author of repo has some time to try to see if changes can be made upstream as this would simplify the patches the port itself must hold onto (a sign of a clean FreeBSD port is the lack of patches not the existence of) but at least kicking the tires on the repo shows it can work for non-nvidia cards (with some effort).