Re: NEW sysutils/openvox-server

Hi Klemens,

On Sun, Feb 1, 2026 at 4:10 PM Klemens Nanni <kn@openbsd.org> wrote:

01.02.2026 15:07, Sebastian Reitenbach пишет:
> Hi Klemens,
>
> same here, tried to address your concerns. Updated tarball attached.
> See inline below.
>
> On Sat, Jan 31, 2026 at 2:58 PM Klemens Nanni <kn@openbsd.org <mailto:kn@openbsd.org>> wrote:
>
>     28.01.2026 16:48, Sebastian Reitenbach пишет:
>     > Hi, 
>     >
>     > cat pkg/DESCR: 
>     > OpenVox is configured in an agent-server architecture, in which a
>     > primary server node manages the configuration information for a fleet of
>     > agent nodes. OpenVox Server acts as the primary server node. OpenVox
>     > Server is a Ruby and Clojure application that runs on the Java Virtual
>     > Machine (JVM). OpenVox Server runs Ruby code for compiling Puppet
>     > catalogs and for serving files in several JRuby interpreters. It also
>     > provides a certificate authority through Clojure.
>     >
>     > It is a drop in replacement for Puppetserver.
>     >
>     > with that, the legacy sysutils/puppetserver/8 port could bite the dust. The version there is loooong time EOL.
>
>     That'd be nice.
>
>     >
>     > between OpenVox server 8.11.x and 8.12.x they removed the Makefile, therefore the old Makefile is taken and stored in the files section. otherwise, could implement it directly in the do-install: section in the ports Makefile. But I found this way easier and a bit cleaner, but if there are concerns, I'd take it on and migrate the target into the do-install: if necessary.
>
>     See my openvoxdb response.  In addition to that
>
>     - MODJAVA_RUN_DEPENDS is not needed in RUN_DEPENDS,
>       the java module already does that
>
>
> removed.
>  
>
>
>     - as mentioned in openvoxserver-ca, I suggest dropping
>       ruby flavors, so the dep here would need adjusting
>
>
> adjusted.
>  
>
>
>     - SUBST_CMD purely on WRKSRC files seems odd, why not
>       do that in post-patch?
>
>
> done, 
>  
>
>
>     This files/Makefile looks simpler and lots of repeated lines
>     could definitely be merged with globbing or .for loops in make,
>     which would result in a perhaps shortish, but still simple
>     and boring do-install.
>
>
> The Makefile is taken from 8.11.0 version distfile and I'm not that well 
> versed with loops in Makefile to update it. Same as in openvoxdb, I'm going
> to check, if they might be open to re-add the Makefile, as it was quite convenient.
>  
>
>     Then we don't need gmake and stuff like 'rundir = /var/run'
>     and 'rm -rf ${PREFIX}/var' would cancel themselves out.
>
>
> I added the ?= to the Makefile, to be able to use the FAKE_FLAGS to 
> override what is in the file. No USE_GMAKE needed.

Same here, looks much better already, but I think some comment explaining
the Makefile situation is needed.

done.

 

OK kn to import, last notes:

- server uses MODJAVA_VER=17+ while db has 17 (no plus)
  is that intentional?

no, not intentional. I usually install DB first, which pulls in right Java version,

I haven't tested with JAVA > 17 yet, so removed the +.

 

- server has rubylibdir=MODRUBY_SITEDIR while db has share/puppetlabs/puppet
  i suppose you run-tested it both ports, but this sill jumps out,
  so just checking

This is by intention. The -plugin subpackage of DB installs the termini, which

in turn has to be copied into the openvoxserver environment, so its jruby can find 

and pick it up. Therefore it's easiest to have it in a separate directory.

 

- in do-install, use INSTALL_DATA_DIR not 'install -d -m 0755'

done.

Thanks a lot for reviewing and feedback, as well as the other openvox server related ports.

cheers,

Sebastian