Early WIP: convert login.conf to UCL
Needs RevisionPublic
Actions

Authored by allanjude on Jun 19 2020, 9:35 PM.

Details

Reviewers

sef
markj
kevans
swills
emaste
michaelo

Summary

This revives a very old project that has been langishing in my local tree

It adds UCL support to cap_mkdb so login.conf can be parsed from UCL (or the original format)

The .db file that is created stays the same, so it doesn't impact compatibility

Diff Detail

Repository

rS FreeBSD src repository - subversion

Lint

Lint Passed

Unit

No Test Coverage

Build Status

Buildable 31836
Build 29396: arc lint + arc unit

Event Timeline

allanjude created this revision.Jun 19 2020, 9:35 PM

Herald added a subscriber: imp. · View Herald TranscriptJun 19 2020, 9:35 PM

allanjude requested review of this revision.Jun 19 2020, 9:35 PM

Harbormaster completed remote builds in B31836: Diff 73355.Jun 19 2020, 9:35 PM

pstef added a subscriber: pstef.Jun 21 2020, 7:52 AM

pstef added inline comments.

usr.bin/cap_mkdb/cap_mkdb.c
89	Is this the BUFSIZ from stdio.h? If so, it's perhaps an unnecessary connotation with setbuf.
138	I'm almost sure that err() and friends terminate output with a newline by definition.
148	I'd suggest `\|\| uclver < 1` because this case doesn't seem to be handled below.
301	A leftover comment?
347	realloc() can return a null pointer
349	a leftover comment?

freqlabs added a subscriber: freqlabs.Jun 21 2020, 10:59 PM

freqlabs added inline comments.

usr.bin/cap_mkdb/cap_mkdb.c
232	Doesn't look like you use i, was it for debugging?
361	Again, i doesn't seem to be used.
usr.bin/login/login.conf
21	Comments about the format will need to be updated.
182	I see a mixture of `foo` -> `foo` and `bar@` -> `bar=true`. Presumably they all are supposed to be booleans set to true? As an aside, I would really like to get UCL to add implicit booleans: https://github.com/vstakhov/libucl/issues/227

mateusz_serveraptor.com added a subscriber: mateusz_serveraptor.com.Jun 24 2020, 7:53 AM

0mp added a subscriber: 0mp.Jul 8 2020, 4:01 PM

Your login.conf changes need to be rebased against an up-to-date version from current - you're undoing the introduction of the "mail" capability and other changes related to environment vars from D21481.

Is it an issue that the login_cap code will now no longer be able to fall back to the text file if the database is not present?

debdrup added a subscriber: debdrup.Feb 12 2021, 8:56 PM

freebsd_igalic.co added a subscriber: freebsd_igalic.co.Feb 12 2021, 8:57 PM

freebsd_igalic.co added inline comments.Feb 12 2021, 9:12 PM

usr.bin/login/login.conf
182	github.com think it's worth being explicit, and instead of having three formats,we should standardise on foo=true

ihor_antonovs.family added a subscriber: ihor_antonovs.family.Feb 23 2023, 4:39 AM

grahamperrin added a subscriber: grahamperrin.Apr 22 2023, 11:57 AM

michaelo requested changes to this revision.Aug 9 2023, 8:57 AM

michaelo added a subscriber: michaelo.

michaelo added inline comments.

usr.bin/login/login.conf
30	Will this change support colon, spaces and commons in env var values? E.g. setenv=BLOCKSIZE=K FOO="bar baz" NO_PROXY="example.com,company.net" HTTPS_PROXY="http://my-proxy.example.com:8000" Especially the last two are crucial for enterprise usage...