This one time, at band camp, William Ballard said:
> On Sat, Feb 19, 2005 at 03:02:00AM +0100, Josselin Mouette wrote:
> > Furthermore, how does a thing being "standard" help the user in his
> > choice? The user only thinks of his own needs, thus a correct
> > wording would be "pick A if you don't care". However the current
> > wording is even better; the question isn't even asked at high
> > priority, and the single file method is silently chosen.
>
> Is the multiple-file configuration logically equivalent to the
> single-file configuration? If you #include'd all the tiny subfiles,
> would the resulting config be identical to the single-file config?
>
> If so, then what are we really arguing about: they are isomorphic.
> Perhaps a tool could generate the single-file config for easier
> double-checking of the split-file config.
>
> If not, then the user needs to know what will behave differently.
The only difference, AFAICT, is that when you pick the split file
option, the split files are concattenated together on /etc/init.d/exim4
{stop,restart,reload} (this is really handled by update-exim4.conf, but
that is irrelevant to the user perspective, IMHO). They produce the
same initial configuration in any case. The only difference from a user
experience is knowing that split files vs. monolithic is really related
to what to edit, rather than where configuration is read at runtime.
--
-----------------------------------------------------------------
| ,''`. Stephen Gran |
| : :' : sgran@debian.org |
| `. `' Debian user, admin, and developer |
| `- http://www.debian.org |
-----------------------------------------------------------------
Attachment:
pgpiOB2NFiTR4.pgp
Description: PGP signature