From mboxrd@z Thu Jan 1 00:00:00 1970 From: Eric Bavier Subject: Re: [PATCH] gnu: Add ghc-semigroups. Date: Thu, 22 Oct 2015 11:54:10 -0500 Message-ID: <4154eab18ab6903b7fbae9dd0fe33a13@openmailbox.org> References: <1445483479-19104-1-git-send-email-ericbavier@openmailbox.org> <1445503671-30014-1-git-send-email-ericbavier@openmailbox.org> <1445503671-30014-15-git-send-email-ericbavier@openmailbox.org> <87mvvbq9a5.fsf@denknerd.org> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:50739) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZpJ8E-0002df-5y for guix-devel@gnu.org; Thu, 22 Oct 2015 12:54:31 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZpJ8A-0006Qb-8K for guix-devel@gnu.org; Thu, 22 Oct 2015 12:54:26 -0400 Received: from smtp6.openmailbox.org ([62.4.1.40]:52294) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZpJ8A-0006Pc-0G for guix-devel@gnu.org; Thu, 22 Oct 2015 12:54:22 -0400 In-Reply-To: <87mvvbq9a5.fsf@denknerd.org> 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.org@gnu.org Sender: guix-devel-bounces+gcggd-guix-devel=m.gmane.org@gnu.org To: Paul van der Walt Cc: guix-devel@gnu.org On 2015-10-22 09:15, Paul van der Walt wrote: > On 2015-10-22 at 10:47, quoth ericbavier@openmailbox.org: >>> + (inputs >>> + `(("ghc-text" ,ghc-text) >>> + ("ghc-hashable" ,ghc-hashable))) >> >> Do these not need to also be propagated? > > This is a comment that comes up frequently. To be honest, i don't know > how to check. My packaging strategy has been, > > 1) put everything in `inputs` > > 2) make sure the package P1 builds > > 3) use P1 as an input elsewhere, say in P2 > > 4) if and only if P2 cannot build because of the usual "broken package, > missing dependencies for P1" error, i add the missing packages to > `propagated-inputs` of P1. > > So no, pragmatically speaking, those inputs do not need to be > propagated, because subsequent builds work fine (up to and including > Idris, which was the root of the dependency tree causing me to add this > and many other packages). Great. Yes, we'd rather not propagate inputs unless strictly necessary. The only reason I ask is for cases where a user might be using the library to build something we do not have packaged, but in that case it's trivial enough them to install the missing dependency. Again, all this should be a moot point with a native-search-path for GHC. -- `~Eric