Notices by Joshua M. Clulow (jmc@m.unix.house)

@whitequark @mei I dunno, I have two degrees and I care probably too much about software quality and I still make programs that do idiot things

@cks I think this is sort of what...

git commit --fixup PREVIOUS_COMMIT

... is for, when then combined with a "rebase --autosquash"?

@lanodan @cks I think 9front probably tolerates a lot more risk (and has a lot less critical deployments) than OpenBSD. I also received the perception that OpenBSD really only care specifically about having no software required to work on the OS shipped from outside their base repository, and they have a pretty firm policy on what licences can be used in there; hence, their work in progress Game Of Trees git reimplementation.

@cks I thought FreeBSD moved to git a few years ago

@cks Yeah I think given they have to rewrite git before they can use it, it might be a while before they switch haha

It is a rich tapestry of feelings to read through the proc(1) manual page from Solaris 2.5, dated literally 30 years ago in 1994:

https://sysmgr.org/~jclulow/sol25/proc.1.html

And then compare it with the modern equivalent in illumos:

https://illumos.org/man/1/proc and https://illumos.org/man/1/ptree

@danmcd Yes! I was looking into the origin of ptree(1), which was present then also. No pgrep(1) or pkill(1) yet, apparently.

@encthenet write(2) to a self pipe you already have open is also safe and well-trodden, in order to inject an event into your event loop. Some OSes also have event loop systems with a facility built for this, such as port_send(3C) from event ports on illumos.

@GeePawHill Definitely a big fan of small monoliths and SQLite and single VMs. Admittedly in AWS because it's been easy, but I am using almost no features of EC2 other than "a small VM please".