bug#25101: [PATCH 1/2] gnu: Add hpaio-enabled sane-backends variant.

[Andy Patterson]

Hi, and sorry for the late reply.

On Tue, 06 Dec 2016 10:10:57 +0100
address@hidden (Ludovic Courtès) wrote:

> Andy Patterson <address@hidden> skribis:
> 
> > On Mon, 05 Dec 2016 21:48:36 +0100
> > address@hidden (Ludovic Courtès) wrote:
> >  
> >> > +(define-public sane-backends+hpaio    
> >> 
> >> Could you add a comment and a synopsis explaining what’s this is?
> >>   
> >
> > Sure, is the following patch ok?  
> 
> Yup!
> 
> >> > +  (package
> >> > +    (inherit sane-backends)
> >> > +    (name "sane-backends+hpaio")
> >> > +    (inputs
> >> > +     `(("hplip" ,(@ (gnu packages cups) hplip))
> >> > +       ,@(package-inputs sane-backends)))    
> >> 
> >> The closure size of this is 290 MiB whereas it’s 87 MiB for
> >> ‘sane-backends’.  I suppose that’s the reason to keep’em separated?
> >>  
> >
> > Yeah, and there's also the fact that hplip depends on sane-backends.
> > But this is the reason I didn't change sane-backends into a private
> > variable, and then make this the new sane-backends.  
> 
> Oh right.
> 
> One last thing: should we renaming “sane-backends” to
> “sane-backends-minimal”, and “sane-backends+hpaio” to “sane-backends”?

The current "sane-backends" isn't really minimal, as it includes most
backends. I'm not sure how it's applied elsewhere in Guix though, so
I'll leave it up to you.

> 
> That way all scanners would work out of the box, at the expense of
> extra disk usage, but that is roughly what patch 2/2 does anyway.

One more patch is required in either case, as hplip would need to use
sane-backends-minimal. Let me know if you'd like me to make this change.

--
Andy