Re: sqlite3

[Eli Zaretskii]

> From: Qiantan Hong <qhong@mit.edu>
> Date: Tue, 14 Dec 2021 17:43:18 +0000
> Cc: "larsi@gnus.org" <larsi@gnus.org>,
>  "emacs-devel@gnu.org" <emacs-devel@gnu.org>
> 
> > We don't yet have a good enough understanding of how useful the
> > current interface is.  Extending and generalizing without any good
> > idea where we want to go is not something I'd like to do in Emacs.
> If we start with the store.el interface, we probably won’t even have
> considered something like multisession.el’s interface.

Or we could find out that it's over-generalized, and the additional
features are not used.  That's what having no experience means: you
can err either way.

> > We just started this experiment, let's see how it unfolds before
> > considering our next steps.
> Putting the more specialized multisession.el interface in master means 
> we will need to provide backward compatibility even if at some point
> store.el is added.

Assuming we decide that extending what we have is indeed necessary;
that is not a given.  And if we do decide, I see no catastrophe: we
are doing this all the time, keeping backward compatibility as we go.
There's no need to be afraid of that.

> I don’t see any downside of the other way — providing just store.el
> interface from the beginning.

Well, I do, and I tried to explain my take of this.  It's okay to
agree to disagree, of course.