On Tue, 19 Jun 2012 08:44:24 +0800, Magnus Hagander <magnus@hagander.net>
wrote:
> On Mon, Jun 18, 2012 at 3:42 PM, Guillaume Lelarge
> <guillaume@lelarge.info> wrote:
>> On Sun, 2012-06-17 at 12:40 -0400, Tom Lane wrote:
>>> Dave Page <dpage@pgadmin.org> writes:
>>> > On Sun, Jun 17, 2012 at 4:39 PM, Magnus Hagander
>>> > <magnus@hagander.net> wrote:
>>> >> Uh, that part is just incorrect. Several sets of platform packages
>>> >> certainly do initdb for you. And AFAIK every single one of them on
>>> >> Linux at least do service setup for you.
>>>
>>> > Hmm, clearly it's been a while since I did a PG installation on
Debian
>>> > (oddly!), as that does seem to leave the server up and running. I'm
>>> > fairly certain it didn't in the past.
>>> > RPMs on the other hand, do not.
>>>
>>> FWIW, Red-Hat-based systems have a strong distro policy against
starting
>>> servers merely because the package got installed --- the theory is
that
>>> an "everything" install should not leave the user running a bunch of
>>> servers he doesn't know about and maybe hasn't configured securely.
>>>
>>> I'm a bit surprised to hear that Debian does it differently; although
>>> it's possible that they distinguish manual from automatic install
>>> scenarios. It's a little bit saner to do an auto service start if you
>>> know that the user explicitly requested this specific package.
>>>
>>
>> When you use aptitude or apt-get to install the server package, it will
>> install the binaires, execute initdb, and start the server. I much
>> prefer the RPM way of doing it (IOW, not starting the server). Anyway,
I
>> guess there are both pros and cons in both methods.
>
> Yeah. But it's a good point that since they are different, I should
> add info to each of the pages saying exactly how it behaves on that
> specific platform.
I'm not sure this is really our responsibility. The package maintainer
should have a webpage explaining how his packages work.
--
Guillaumehttp://www.postgresql.frhttp://dalibo.com
