.gaim Revisited

[Luke Schierer]

On Wed, Mar 21, 2007 at 08:18:36AM -0500, Richard Laager wrote:
> On Wed, 2007-03-21 at 08:51 -0400, Ethan Blanton wrote:
> > Richard Laager spake unto us the following wisdom:
> > > Earlier, Sean and I were discussing what to do about .gaim, rename-wise.
> > > He said he was thinking about following the XDG Base Directory
> > > Specification:
> > > http://standards.freedesktop.org/basedir-spec/basedir-spec-0.6.html
> > 
> > I saw this spec before, I saw you mention it, I thought it was a
> > horrible idea, and I hoped that this had died.  I see that it has not.

I agree with all of Ethan's reasoning against this spec.  I'd love to
see something help organize ~/, but as I see it, the primary problem is
not the zillions of ~/.appname directories that can happen, but the fact
that some applications (*ahem* Gnome) don't actually have everything
there.  Thus I have .gnome, .gnome2, .gnome_private, .gnome2_private,
.gconf, .gconfd, and a couple others for application specific stuff that
somehow didn't find its way into one of those.

> 
> Blame Sean. ;)
> 
> Perhaps we just need some migration code to move .gaim to .purple and we
> can call it good? Or should we just use .purple if it's there and if
> not, fall back to .gaim if it exists?

As I see it, a continued use of .gaim will make sense to a vast majority
of our users.  It will however make increasingly less sense as time goes
one.

1)new users will be utterly confused, because they won't realize that we
were called gaim at one point.
2)even some of our existing users will eventually forget the connection
here

using .purple seems to make sense, it would allow for any UI to store
data in the same place, data that you probably want to be read by any UI
(logs, accounts, states, so on).  It is a strongly tempting idea.

But I think ultimately it must be rejected.  Just as having some stuff
in .gimp and other stuff in .gconf is confusing, having stuff in .purple
will be confusing when I'm running pidgin.  It will be particularly
confusing if I'm a relatively clueless user who is right now today
having trouble with the concept that .gtkrc-2.0 is involved with gaim.

Similarly, while it is convieable that at some point a given user might
run pidgin in linux and robin in win32, and want to share data, it is
less concievable that many users besides us will want to run both pidgin
and finch.  

Those who do fit that category will be, almost inherently, that set of
users who can handle creating symlinks to do the job of unifying the
two.

luke