[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#926306: RFS: socklog/2.1.0-9



On Wed, Apr 10, 2019 at 08:20:30AM +0000, Dmitry Bogatov wrote:
>
> [2019-04-08 01:03] Mathieu Mirmont <mat@parad0x.org>
> > Done, the init scripts call daemon(1) and runsv(1) and they work
> > pretty nicely.
>
> Are you sure you need "Depends: runit"? Maybe it is just me, but I
> thought that `socklog' is just an implementation of `syslog'. If
> so, hard dependency on bin:runit is quite heavyweight.

It heavily relies on runsv and svlogd, that's how it splits files by
syslog service name. Essentially socklog is a data collector for svlogd.

The alternative would be to re-implement runsv and svlogd by redirecting
the output of socklog to a file and using logrotate, but clearly that
would be a hack, it's not how it's meant to be run, and it would have
the risk of losing messages when logrotate kicks in.

The dependency on runit is less risky than that. Also the dependency is
on runit alone, not on any of the runit-* packages that call runsvdir,
so it's quite lite.

> > There is one more issue that I noticed this weekend: the orig.tar.gz
> > file that is registered in debian archives is not the same as the
> > upstream tarball. It is in fact a tarball of the upstream tarball
> > (!). I don't know why it's done this way, and it pretty much breaks
> > breaks source format 3.0 (quilt) because I can't get dpkg-source to
> > unpack the tarball before applying the patches. Do you know how to
> > deal with that?
>
> You can repack it as new upstream version. New version would be
> something like `2.1.0+repack-1'. Do not forget add clarification into
> Debian.source.

Ah! You made my day, I didn't know this was possible!

-- 
Mathieu Mirmont <mat@parad0x.org>

Attachment: signature.asc
Description: PGP signature


Reply to: