Re: why etc/default.d & etc/local.d

[Noah Slater <ns...@apache.org>]

The premise of this question is flawed. Debian is a downstream. So are all
distributions. CouchDB is designed, first and foremost, to package itself
from source. That's why we create these directories for the user. Please
also note that they are only created during "make install" and they do not
exist in our source.


On 13 November 2012 00:43, Randall Leeds <ra...@gmail.com> wrote:

> On Mon, Nov 12, 2012 at 3:20 PM, Jan Lehnardt <ja...@apache.org> wrote:
>
> >
> > On Nov 13, 2012, at 00:15 , Paul J Davis <pa...@gmail.com>
> > wrote:
> >
> > > I'd rather leave them or remove the functionality. Hiding the config
> > chain seems wrong.
> >
> > +1 on definite decisions, and +1 on keeping things as is.
> >
>
> The functionality is to pass a config directory via the command line with
> the -A switch.
>
> It would be the sort of thing where the debian package would have
> COUCHDB_OPTIONS="-A /etc/couchdb/default.d -A /etc/couchdb/local.d" in
> /etc/default/couchdb. On RHEL it would be /etc/sysconfig/couchdb.
>
> The change would be that running the couchdb start script by hand (not
> using the init script) would probably no longer assume the existence of
> these directories nor pass the equivalent switch implicitly.
>
> Again, I'm +0, but just wanted to make sure we were clear on what's going
> on. The "-A" flag is not in question. Only the implicit -A on these two
> directories and the presence of these (empty) directories in our source
> tree.
>



-- 
NS

Re: why etc/default.d & etc/local.d

[Benoit Chesneau <bc...@gmail.com>]

On Sun, Dec 2, 2012 at 3:14 AM, Noah Slater <ns...@apache.org> wrote:

> The premise of this question is flawed. Debian is a downstream. So are all
> distributions. CouchDB is designed, first and foremost, to package itself
> from source. That's why we create these directories for the user. Please
> also note that they are only created during "make install" and they do not
> exist in our source.
>
> Well they actually exists:

https://github.com/apache/couchdb/tree/master/etc

even if they are templates. I am not sure we need to have this init and
such created by couchdb since it is the responsibility of the maintainer
and only works on some distribution. Not a big deal anyway but since they
won't work on all distributions, i am not sure i see the point to have them
during install.

- benoit



> On 13 November 2012 00:43, Randall Leeds <ra...@gmail.com> wrote:
>
> > On Mon, Nov 12, 2012 at 3:20 PM, Jan Lehnardt <ja...@apache.org> wrote:
> >
> > >
> > > On Nov 13, 2012, at 00:15 , Paul J Davis <pa...@gmail.com>
> > > wrote:
> > >
> > > > I'd rather leave them or remove the functionality. Hiding the config
> > > chain seems wrong.
> > >
> > > +1 on definite decisions, and +1 on keeping things as is.
> > >
> >
> > The functionality is to pass a config directory via the command line with
> > the -A switch.
> >
> > It would be the sort of thing where the debian package would have
> > COUCHDB_OPTIONS="-A /etc/couchdb/default.d -A /etc/couchdb/local.d" in
> > /etc/default/couchdb. On RHEL it would be /etc/sysconfig/couchdb.
> >
> > The change would be that running the couchdb start script by hand (not
> > using the init script) would probably no longer assume the existence of
> > these directories nor pass the equivalent switch implicitly.
> >
> > Again, I'm +0, but just wanted to make sure we were clear on what's going
> > on. The "-A" flag is not in question. Only the implicit -A on these two
> > directories and the presence of these (empty) directories in our source
> > tree.
> >
>
>
>
> --
> NS
>