Re: [siesta-dev] Config committals

[prev] [thread] [next] [lurker] [Date index for 2002/09/03]

From: Simon Wistow
Subject: Re: [siesta-dev] Config committals
Date: 22:19 on 03 Sep 2002
On Tue, Sep 03, 2002 at 07:22:52PM +0100, Richard Clamp said:
> The observation: Now we have two different ways of accessing the list
> database, one via Siesta::Storage (which is used by the existing
> Siesta::List and Siesta::User code), and another via the
> Siesta::Config hierarchy.

This is true. And it's why I bought it up a while back. But I thought
'fuck it, we need key value config type stuff and I'm fucking bored and
I need something to do in between editing footage of ECTS and
completing Max Payne'. Sorry if that sounds snippy but I was struck with
a sudden attack of JFDI.

Basically - as far as I can see it anyway - we do need two types of
storage which are objects (lists, users etc) and config (key and

I agree with Greg about the three different tables thing - that's, err,
why I bought it up. Although there are problems with it based on the
fact that per-user-per-list config needs list and user references. 

I'm not sure what to do - in some way I'm with Jwz on this one

"Now wait, I see two different problems: you need a mail summary file;
and you need a representation for address books. It's not obvious to me
that those are anywhere *near* the same problem, or that one should try
to conflate them and solve them simultaniously. The requirements of the
two problems seem pretty dramatically different to me. 

On the one hand, it's a drag to do two different implementations, but on
the other hand, it's a drag to have one of the two problems be solved
badly because of baggage from the other problem; or to have the
all-singing-all-dancing solution take longer than two independent
solutions together."

The problems of storing Plugin config is not that same as extracting
Users and Lists from a database. 

Whether the entry point for the two should be the same is a different
matter but that's easy enough to change.

> Some of those tests rapidly fail because the new stuff relies on your
> file existing, rather than using the one created by
> t/05setup_database.t

I'll fix this anon.


There's stuff above here

Generated at 13:56 on 01 Jul 2004 by mariachi 0.52