Re: questions about database installations
On 02/08/2012 02:12 AM, Paul Wise wrote:
> On Wed, Feb 8, 2012 at 1:13 AM, Nikolai Lusan wrote:
>> Last year I packaged cluebringer (version 2.0 of policyd for doing
>> greylisting and other anti spam controls) as postfix-cluebringer. In my
>> initial packaging I used dbconfig-common to handle the database
>> installation and configuration, however when I did an RFS the sponsor
>> pulled it all out. I specifically put those things in place so that the
>> installation of the database and the configuration would be simpler for
>> end users (being done through debconf).
>> I would like to keep the dbconfig-common method (even though getting it
>> up and running was a pain) simply because it is easier for end users to
>> get the whole thing up and running, and it provides and easier method
>> for updating the database as need be. Do people think that this is a
>> good idea? Or am I better off looking for another method to handle this?
> In general it is a good idea, I'm not able to judge this case.
> Toni, could you explain your reasons for removing the dbconfig-common stuff?
Of course I can. I have just forwarded you (Paul) the mail exchange
pertaining to that case that should shed some light on the initial
problems, but the gist of it is that I could not get the dbconfig-common
stuff to run reliably, while the freeze was approaching rapidly. As I
wanted the package to really make it into Squeeze, Nikolai was mostly
unavailable to help at that time, and I was stuck, I removed everything
that I could not fix, thus making at least the leftovers work correctly,
and the package move into the release.
There's generally nothing wrong with using the available toolchain, but
a working bare-bones package takes precedence over a non-working package
with bells and whistles, in my book at least. I can also provide a
bazaar or git repo that details the evolution of the package, if
desired. That would allow you to form your own opinion, and maybe
provide me with feedback about what went wrong.