12:02 <farkuhar> not sure if it's the same for the shadow source tree that was released yesterday, but the version in ports has the following in src/useradd.c, line 82: "static gid_t def_group = 1000;" This is the new default when /etc/default/useradd does not exist, despite what the man-page says.

13:38 <farkuhar> jaeger: http://ix.io/482x successfully builds the latest version of shadow, and should address the oddity you noticed with useradd. Also might want to delete the outdated patches in the ports tree.

14:05 <beerman> why not deploy a defaults file?

14:22 <farkuhar> in the *BSD world, it's considered a bug when the man page disagrees with the observed behaviour. Deploying a defaults file only hides the discrepancy, rather than eradicating it at the source.

14:25 <farkuhar> to actually eradicate the discrepancy at the source, we would have to file a bug report with the upstream maintainers of shadow. Being a Linux project, they might not care as much about the man-page getting out of sync with the observed behaviour.

14:30 <stenur> shadow-4.11.1:libmisc/find_new_gid.c does not have any USE_PAM

14:37 <stenur> login.defs should be inspected unless you create a system group

15:00 <beerman> man 8 useradd actually reports that GID_MIN is 1000, which looks like what you are changing with your sed, the def_group

15:00 <beerman> its the same for UID_MIN

15:00 <beerman> https://github.com/shadow-maint/shadow/commit/bbf4b79bc49fd1826eb41f6629669ef0b647267b

15:02 <beerman> the commit in question is almost 2 years old. they didn't change the man file which means to me that it looks like they got rid of that bug. it does say under the explanation for the -g switch that its still 100, so thats a problem. but it seems like they are changing the default to start with 1000 instead of a 100

15:03 <farkuhar> yes, they forgot to change the man page in the section that explains the -g switch, and that's what my sed was trying to follow. Reading further it seems like a better solution is to change the paragraph about -g

15:05 <farkuhar> but it still doesn't address the fact that /etc/group doesn't have any entry corresponding to gid 1000, which is what jaeger initially found so puzzling.

15:06 <beerman> yeah, true. it looks like yet another of CRUXs oversights over the years

15:10 <stenur> well actually it should get created alongside the user shouldn't it?

15:11 <farkuhar> easiest way to reconcile the differences? change our core/filesystem port so that /etc/group has users::1000, and run sed -i '/100 by default/ s/100/1000/' man/man8/useradd.8 before building core/shadow?

15:12 <farkuhar> stenur: depends on the setting for USERGROUPS_ENAB in login.defs

15:12 <stenur> ah ja now i see. useradd.8 has not been updated, the code is all 1000, everywhere

15:12 <stenur> ..and must have been a long time, as _i_ have gid 1000, and i installed the thing in 2019

15:13 <stenur> na was interactive, farkuhar

15:13 <stenur> but that seems true nonetheless, for defaults

15:13 <stenur> oh. gcc 12.2 has just been released.

15:15 <stenur> "first bug-fix release", "important fixes for regressions and serious bugs in GCC 12.1 with more than 71 bugs fixed since the previous release".

15:15 <stenur> production time ready gcc 12 ;->

17:06 <farkuhar> http://ix.io/483H is the minimal git diff to implement my last suggestion.

21:04 <beerman> farkuhar: https://github.com/shadow-maint/shadow/pull/553 fyi

22:10 <farkuhar> beerman: thanks!