[bug#28170] Add gnutls/dane + use it where its needed (gnurl, libmicrohttpd, gnunet)

[Christopher Baines <mail@cbaines.net>]

[-- Attachment #1: Type: text/plain, Size: 1657 bytes --]

On Sat, 30 Sep 2017 12:23:09 +0000
ng0 <ng0@infotropique.org> wrote:

> Christopher Baines transcribed 2.1K bytes:
> > On Mon, 21 Aug 2017 09:57:26 +0000
> > ng0 <ng0@infotropique.org> wrote:
> >   
> > > The dependency chain of GNUnet demands GnuTLS with DANE support.
> > > You can use it without DANE, but there are certain parts which
> > > will not work. DANE is recommended. (ports of FREEBSD uses
> > > gnutls-dane for libmicrohttpd aswell for example).  
> >   
> > > From c9dedfd1a6f91d557006b00505a428be84102b1e Mon Sep 17 00:00:00
> > > 2001 From: ng0 <ng0@infotropique.org>
> > > Date: Mon, 21 Aug 2017 09:28:51 +0000
> > > Subject: [PATCH 1/4] gnu: gnutls: Add 'gnutls-dane'.
> > > 
> > > * gnu/packages/tls.scm (gnutls/dane): New variable.  
> > 
> > Hey, so I'm trying to do some reviewing.
> > 
> > What was your reasoning for creating a new gnutls/dane package,
> > rather than adding unbound to the existing gnutls package?
> > 
> > I don't know much about GnuTLS, but it would be good to make it
> > clear why this approach was taken, especially in the commit message
> > and by the gnutls/dane package definition.  
> 
> Honestly, I did not want to touch the gnutls package for this.
> I wanted to leave the option open to have gnutls without dane.
> Sure, it's a useful feature. If you think I should apply it
> directly to gnutls, tell me and I'll send a new patch.

I'm happy with that reason, if you could put that in the relevant
commit message, or in a comment by the gnutls/dane package definition
(or both), I'll check I can build all the changed packages, and push
if they all build for me.

[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 963 bytes --]