Re: FWIW: script vs. configurtion file
On 7/24/14, saint@eng.it <saint@eng.it> wrote:
> Joel Rees writes:
> > On Wed, Jul 23, 2014 at 9:29 PM, <saint@eng.it> wrote:
> > > Zenaan Harkness writes:
> > >
> > >> So thank you Joel for spending the time to describe these
> > > > concepts as 'pedantically' as you have. Your descriptions
> > > > are an excellent grounding for the conversation which is
> > > > undoubtedly going to continue :)
> > >
> > > One question. Can you give me an example of Turing completeness with
> > > just declarations?
> > >
> > > If not, the description you refer to may be valid only when discussing
> > > "configuration", and even there there is some loss.
> 2) When a program has a declarative configuration file, then you can
> select among a certain number of fixed behaviours. When a program
> has a Turing complete language for its configuration then that
> program is likely to be quickly extensible.
>
> I admit that you could write a declarative configuration that lets
> you specify "scripts" to be executed by the "configured" program
> or the system. Frankly, I would put everything into a "script", much
> easier to write and maintain.
Each to his own thank you! Yes, taste it is :)
Scripting has its place, but from my extensive reading
of systemd docs and some of the old sysv startup scripts
(for postfix and various others over the years), give me
systemd unit files any day! Preference. Mine.
For games customization, scriptability sounds to make
a lot of sense to me. Might also make sense for some system
daemons, I dunno. But where a bunch of flags can be
specified, and satisfy typical/all daemon config needs,
declarative unit files appear *much* cleaner to my simpleton
eyes.
Time will tell I guess, but I strongly suspect that in daemon
land, unit files will prove to very successfuly cover the majority
of configuration use cases, with only rare/occasional call outs
to "user authored" scripts. Just got a gut feeling on this one.
Regards,
Zenaan
Reply to: