From mboxrd@z Thu Jan 1 00:00:00 1970 From: Leo Famulari Subject: Re: [PATCH] gnu: Add capnproto. Date: Fri, 19 Aug 2016 18:06:24 -0400 Message-ID: <20160819220624.w5wu4c47ltzpud4u@jasmine> References: <87ziog1p4r.fsf@ike.i-did-not-set--mail-host-address--so-tickle-me> <20160814171756.GA27717@jasmine> <87wpjjgqrt.fsf@ike.i-did-not-set--mail-host-address--so-tickle-me> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:49764) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1barvy-00088C-HC for guix-devel@gnu.org; Fri, 19 Aug 2016 18:06:43 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1barvt-0002WF-Rs for guix-devel@gnu.org; Fri, 19 Aug 2016 18:06:38 -0400 Received: from out2-smtp.messagingengine.com ([66.111.4.26]:53064) by eggs.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1barvs-0002V6-LR for guix-devel@gnu.org; Fri, 19 Aug 2016 18:06:33 -0400 Content-Disposition: inline In-Reply-To: <87wpjjgqrt.fsf@ike.i-did-not-set--mail-host-address--so-tickle-me> 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" To: Marius Bakke Cc: guix-devel@gnu.org On Sun, Aug 14, 2016 at 07:16:22PM +0100, Marius Bakke wrote: > Leo Famulari writes: > > > On Sat, Aug 13, 2016 at 07:48:36PM +0100, Marius Bakke wrote: > >> Note that it bundles googletest; I tried unbundling but it proved > >> difficult. gtest will no longer be used from the 0.6 release so I did > >> not think a comment was necessary. > > > > I am cc-ing Lukas with my reply, since he added our googletest package. > > Maybe he has some insight. > The problem is that their autotools system expects to build gtest as > well, so there are references all over. See: > https://github.com/sandstorm-io/capnproto/blob/release-0.5.3/c%2B%2B/Makefile.am > and also configure.ac. It would have to be patched out, which seems > excessive for a 4MB build dependency. In my opinion, it's generally not about the size of the bundled dependency. Rather, it's about the opacity of the dependency graph of the application that does the bundling. The worst case would be something like a bundled OpenSSL, for example. In this case, Debian has accepted the bundled gtest [0], which makes me wonder if my understanding of gtest is incorrect. Perhaps it is designed to be bundled? [0] https://packages.debian.org/sid/devel/capnproto