daemon: add long_opts

[ihor_antonovs.family]

Summary

Long options improve readability of scripts, makes code comprehension easier.
This patch adds long options while preserving existing CLI interface.
Also --help/-h option is added

[ihor_antonovs.family]

ready for review

[pauamma_gundo.com]

usr.sbin/daemon/daemon.8

82–83 ↗

(On Diff #115984)

Should the defaults be given both here and in the individual flag descriptions?

89–91 ↗

(On Diff #115984)

I'm not sure whether here and elsewhere in DESCRIPTION where options (other than the one being described, ie -o --output-file here) are mentioned, it's better to use the long form or the short form as diff(1) and others do. Opinions welcome.

158 ↗

(On Diff #115984)

s/syslog tag/syslog-tag/

[ihor_antonovs.family]

usr.sbin/daemon/daemon.8

89–91 ↗

(On Diff #115984)

IMHO long form improves readability everywhere, I don't have to look up what -H means every time. Short options are difficult to remember unless you work with this tool every day.

But that being said I am happy to revert this if other reviewers feel the same.

[kevans]

Meta: please generate phab diffs with full context included (-U999999, just really high); it makes reviewing a bit easier in the average case where additional context in the file might affect the review process (e.g., checking for consistency)

usr.sbin/daemon/daemon.8

60 ↗

(On Diff #115984)

Prevailing style is to insert a comma between the short option and long option in these listings, e.g.,

.It Fl c , Fl -change-dir

usr.sbin/daemon/daemon.c

86

Might as well const this

134

I think we can drop this line, it's generally assumed that one will seek out the manpage for details.

[ihor_antonovs.family]

address comments by @kevans

[ihor_antonovs.family]

missing closing brace

[ihor_antonovs.family]

minor fix

[ihor_antonovs.family]

@kevans: fixed all comments, added diff with -U9999999

[pauamma_gundo.com]

Manual page changes LGTM. Can't attest to source code or consistency between them.

[freebsd_igalic.co]

how about --help?

usr.sbin/daemon/daemon.c

134

why not add a --help while you're at it?

this way, drop the exit(1) from usage(), and exit as appropriate from main().

[ihor_antonovs.family]

usr.sbin/daemon/daemon.c

134

I was tempted to do this, but I was not sure if reviewers will like it. Not everybody likes the scope creep. I was planning to do this in a separate revision, but I will gladly add it to this diff!

[ihor_antonovs.family]

Add --help as per @freebsd_igalic.co request

[freebsd_igalic.co]

usr.sbin/daemon/daemon.c

111

how we could pass on an option here if we're printing onto stderr or stdout, because often you'd want to foo --help | grep topic

but i'd leave this to more experienced people

[ihor_antonovs.family]

usr.sbin/daemon/daemon.c

111

I like the idea and this is easy to do, but I will keep this feature for the next diff, I want to keep this change scoped

[allanjude]

usr.sbin/daemon/daemon.c

86

l: is specified twice

234

without the braces around this if, this exit(1) runs unconditionally

[imp]

usr.sbin/daemon/daemon.c

221

usage(0) would be better, and adjust usage to take a parameter and exit that value. Avoids errors like no {} from creeping in.

[allanjude]

usr.sbin/daemon/daemon.c

221

and could double for the other requested change, 'if usage() is not an error, write to stdout instead of stderr'

[ihor_antonovs.family]

usr.sbin/daemon/daemon.c

86

It is not me, it was like that before, I promise :)

Will fix.

221

ack, will do

234

ah, my bad , that was late change after adding --help...

[imp]

usr.sbin/daemon/daemon.c

221

I'm not aware of another program does that in the tree, so I'd suggest that we don't do that here.

[ihor_antonovs.family]

Address @allanjude and @imp comments

[ihor_antonovs.family]

minor fix

[allanjude]

Reviewed by: allanjude

[ihor_antonovs.family]

add __builtin_unreachable(); for --help case
remove unnecessary usage() prototype