On Wed, Apr 27, 2016 at 01:05:49PM -0400, Leo Famulari wrote:
> On Wed, Apr 27, 2016 at 12:00:59AM +0300, Efraim Flashner wrote:
> > On Tue, Apr 26, 2016 at 01:13:09PM -0400, Leo Famulari wrote:
> > > On Tue, Apr 26, 2016 at 04:33:16PM +0200, Ricardo Wurmus wrote:
> > > > 
> > > > Hi Efraim,
> > > > 
> > > > > On a whim I removed gtk2 and glib from pinentry and the size of the closure
> > > > > dropped from 700 MiB to 100 MiB. While I was at it I built a version with
> > > > > qt5, if someone wanted it to match a future kde/qt-based desktop. I couldn't
> > > > > get a gtk3 version building, but I didn't try too hard.
> > > > >
> > > > > Efraim Flashner (2):
> > > > >   gnu: Add minimal pinentry.
> > > > >   gnu: Add pinentry-qt.
> > > > >
> > > > >  gnu/packages/gnupg.scm | 34 +++++++++++++++++++++++++++++-----
> > > > >  1 file changed, 29 insertions(+), 5 deletions(-)
> > > > 
> > > > These both look good to me.  I think our policy requires to split the
> > > > first patch in two, one to reduce pinentry, the other to add
> > > > pinentry-gtk2, but I’m not sure.
> > > 
> > > I had the same thought, but it would mean we have a commit in our
> > > history where we *don't* have the GTK pinentry. I think we should avoid
> > > that.
> > 
> > I thought about that too. With the patches I have we'll have people who
> > have pinentry installed who suddenly find that they don't have a gtk-2
> > pop-up anymore. I suppose another option would be to leave pinentry its
> > name and call the curses-only pinentry pinentry-curses or
> > pinentry-minimal or pinentry-tty.
> 
> Good point. That would be a poor experience for our users.
> 
> Although I would prefer if the console-based pinentry was called
> pinentry, I'd rather not change the meaning of the existing variable for
> a program that is so important.
> 
> So, pinentry-tty or pinentry-curses makes sense to me.

I'll double-check the descriptions to add that the gtk/qt versions also
have the curses interface, don't want people to think they need
pinentry-curses and pinentry-gtk.

> 
> > In terms of the commit order, maybe one to rename pinentry->pinentry-gtk
> > and a second to add pinentry?
> 
> Since I think it's best if users of 'pinentry' don't experience any
> disruption, perhaps we could add a variable pinentry-gtk that inherits
> from pinentry? Maybe eventually we could deprecate 'pinentry'.

I like that idea.

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted