[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Bug#61308: PROPOSAL] Initializing databases by using conffiles.

On Wed, 29 Mar 2000, Anthony Towns wrote:

> On Wed, Mar 29, 2000 at 02:10:54PM +0200, Santiago Vila wrote:
> > \begin{proposal}
> > Do not initialize a text database by using the conffile mechanism.
> > \end{proposal}
> What is a `text database' ? That term seems very unclear to me; but
> based on the bug report I'd certainly be happy to second something
> clearer to achieve this.

By text database I refer to a database which is in plain text format,
like /etc/printcap, /etc/passwd, /etc/group or /etc/mailcap.

Certainly, a binary database should not be initialized by using the
conffile mechanism either (for the same reasons), so probably the
word "text" is not needed here.

> Basically, the idea seems to be `if there's an update-<foo> program,
> the file that's being updated shouldn't be a conffile'. Yes?

Well, this raises the following issue: How many ways to update a database
are there? (update-<foo> is one way, but there are some others, NIS uses a
Makefile, for example). Should a database will be allowed to be a conffile
if it is updated by a method different than update-<foo>? I would like
to see an example of database for which the conffile mechanism is useful.


 "279b3a6f5171d5690789f023234c36d0" (a truly random sig)

Reply to: