On Fri, Oct 12, 2012 at 12:38PM, Jos Backus wrote:
> On Fri, Oct 12, 2012 at 1:34 AM, Steve Loughran <[email protected]>wrote:
> 
> > On 12 October 2012 00:27, Jos Backus <[email protected]> wrote:
> >
> > >
> > > TL;DR: init scripts should die.
> > >
> > >
> > +1, except: s/init/r/bash/
> >
> 
> If you are trying to say that shell scripts are a lousy way to write
> scripts that handle complex configurations, I agree. It would be much
> better if all this data currently residing in dozens of shell/environment
> variables was stored in a single YAML file, but shells can't handle that.
> I'm likely to go with an envdir to store this data. Or maybe I'll write
> some Ruby code.

(some clearly provocative questions)

May I ask why dog-slow Ruby is better than a shell? ;) Especially, why it is
better in the frame of a Java-based stack?

Also, I'd like to comment on the topic of the YAML, which sucks badly as any
other generalized serialization formats (circa SGML) compare to DSLs:)

Cos

> But I also think the SysV derived startup mechanism needs to be replaced.
> Pidfiles are a bad solution to a problem that's already well supported by
> the process table, fork()/exec*() and wait*(). Those who don't understand
> UNIX are doomed to reimplement it - poorly.

Attachment: signature.asc
Description: Digital signature

Reply via email to