From mboxrd@z Thu Jan 1 00:00:00 1970 From: zimoun Subject: Re: Build systems and implicit inputs Date: Tue, 21 Jan 2020 19:02:18 +0100 Message-ID: References: <8736ldq74z.fsf@netris.org> <20190719202906.lbanx5puk7t6q4cr@cf0> <87a7753boq.fsf@ambrevar.xyz> <87h813wah0.fsf@gnu.org> <87v9piut40.fsf@ambrevar.xyz> <87o8v5ukgb.fsf@ambrevar.xyz> <87d0blhr9s.fsf@ambrevar.xyz> <20200116190644.uytvzvypuvdwh2iq@n0> <871rry10ow.fsf@ambrevar.xyz> <87o8uxdqnh.fsf_-_@gnu.org> <87o8uxq7ph.fsf@ambrevar.xyz> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:470:142:3::10]:51378) by lists.gnu.org with esmtp (Exim 4.90_1) (envelope-from ) id 1itxrI-0001zv-9m for guix-devel@gnu.org; Tue, 21 Jan 2020 13:02:37 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1itxrH-0003tm-5o for guix-devel@gnu.org; Tue, 21 Jan 2020 13:02:36 -0500 In-Reply-To: <87o8uxq7ph.fsf@ambrevar.xyz> List-Id: "Development of GNU Guix and the GNU System distribution." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-devel-bounces+gcggd-guix-devel=m.gmane-mx.org@gnu.org Sender: "Guix-devel" To: Pierre Neidhardt Cc: Guix Devel Hi Pierre, On Tue, 21 Jan 2020 at 14:07, Pierre Neidhardt wrote: > zimoun writes: > > On Tue, 21 Jan 2020 at 11:56, Ludovic Court=C3=A8s wrote= : > >> > The solution to your problem in my opinion is simply to expose just = the > >> > right amount of options through #:arguments for all build systems. > >> > Would that be satisfactory to you? > >> > >> I think the issue of tweaking the build system and its implicit inputs > >> must be addressed separately. We first need a good API to do that. > >> When we have it, it=E2=80=99ll be nice and easy to drive it via packag= e > >> parameters. :-) > > > > Now I have a better understanding about "package parameters", I agree > > that it is 2 separate stories. > > Hmm... but does it have to? It seems to me that we would gain a lot in > keeping those parameters general enough and not separate the handling of > the build system from the rest. It would be simpler and more powerful. I do not have a strong opinion. From my understanding, how to pass arguments to the build system is a story and how to pass arguments to packages is another one. Roughly speaking, they do not refer to the same record, to the same functions that digest them, etc.. And I am not a fan of adding global variables here and there. Well, I am almost sure that treat them together will end with a big mess. := -) Cheers, simon