From: Mike Larkin Subject: Re: vmd: remove PROC_CONTROL bits from proc.[ch] To: Martijn van Duren Cc: tech@openbsd.org, Dave Voutila Date: Tue, 18 Nov 2025 23:33:21 -0800 On Mon, Nov 17, 2025 at 10:26:24AM +0100, Martijn van Duren wrote: > On 11/17/25 03:25, Mike Larkin wrote: > > On Fri, Nov 14, 2025 at 10:07:46AM +0100, Martijn van Duren wrote: > >> Hello all, > >> > >> A long time ago I removed the control bits from snmpd's proc.c, because > >> snmpd lost any use for snmpctl. Revisiting proc.c made me realize that > >> control setup functionality doesn't need to be in proc.c, and only > >> forces us into creating the process, regardless of being needed. > >> > >> Since I have the impression that vmd is currently the most actively > >> maintained proc.c consumer I started here, but it should probably go > >> just as well for the other consumers, which I'll visit if people agree > >> with me here. > >> > >> Diff below removes all the control bits from proc.[ch] and places them > >> in their equivalent places inside control.c and struct vmd. Since vmd > >> doesn't allow for multiple control sockets I've only took over the > >> main, and dropped the TAILQ bits. > >> > >> Thoughts? OK? > >> > > > > dv@ should decide here. not sure I grok this but at first read it seems like > > shuffling deck chairs? what benefit does this gain? > > It is shuffling deck chairs. The short: maintainability of proc.c in > general. > The PROC_CONTROL bits add special casing and complexity to an already > complex file without any additional value. Removing this special > casing makes proc.c more readable, and when also applied to the other 3 > proc.c daemons (iked, httpd, relayd) makes them that little bit more in > line with eachother.> > >> martijn@ > dv@ should make the call here.