Re: Complicated re-distribution of pgjdbc the "open source way"

On 8 March 2016 at 20:41, Pavel Raiskup <praiskup(at)redhat(dot)com> wrote:

I appreciate any work done the open source way.
>

I take issue with that (in the context of this conversation).

I think you are conflating "the Fedora / Debian free-software purist Linux
distro packaging way" with "the open source way".

There is not one "open source way". There are many. Yours is not the only
one, and I think you're getting a few people off-side by refusing to
acknowledge that just because you want to do things a particular way that
doesn't make your way inherently the only correct one and everyone else's
wrong. Or maybe I'm just reading your tone wrong and being too prickly, but
offlist discussion confirms to me that I'm not the only one if so.

Maven is open source. Waffle is open source. JBoss AS is open source.
OpenOffice is open source, yet causes a lot of packaging pain too. Perl is
open source, but CPAN makes packagers want to cry (I know, I've had to
package CPAN modules). Ruby is open source, but the RubyGems system's
interaction with packaging can be agonizing.

Packaging can be painful. Sometimes that's due to people doing insane
things in the software you're trying to package. Sometimes that's due to
necessary-but-frustrating-and-limiting distro/packaging policies. Sometimes
it's due to more fundamental differences in the models of how distro
packaging works vs how some other software works. If it's hard to package
that doesn't make it wrong.

I think the way forward here is to propose a build profile set up that lets
you omit the things you want to omit, including tests showing that the
omitted features fail in a well-specified and graceful way. I at least have
no objection to running >1 build configuration in TravisCI etc.

--
Craig Ringer http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services