From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:44181) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1eCLiQ-0003dG-SP for guix-patches@gnu.org; Wed, 08 Nov 2017 03:28:08 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1eCLiM-0006S6-Gq for guix-patches@gnu.org; Wed, 08 Nov 2017 03:28:06 -0500 Received: from debbugs.gnu.org ([208.118.235.43]:48853) by eggs.gnu.org with esmtps (TLS1.0:RSA_AES_128_CBC_SHA1:16) (Exim 4.71) (envelope-from ) id 1eCLiM-0006Rr-Cl for guix-patches@gnu.org; Wed, 08 Nov 2017 03:28:02 -0500 Received: from Debian-debbugs by debbugs.gnu.org with local (Exim 4.84_2) (envelope-from ) id 1eCLiL-00049A-SR for guix-patches@gnu.org; Wed, 08 Nov 2017 03:28:01 -0500 Subject: bug#29181: [PATCH 1/1] gnu: mesa: Use symlinks instead of hard links. Resent-To: guix-patches@gnu.org Resent-Message-ID: From: ludo@gnu.org (Ludovic =?UTF-8?Q?Court=C3=A8s?=) References: <20171106214057.12590-1-ludo@gnu.org> <20171106214252.12702-1-ludo@gnu.org> <87r2t93npj.fsf@fastmail.com> Date: Wed, 08 Nov 2017 09:27:29 +0100 In-Reply-To: <87r2t93npj.fsf@fastmail.com> (Marius Bakke's message of "Tue, 07 Nov 2017 20:15:04 +0100") Message-ID: <87k1z1w4y6.fsf@gnu.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: guix-patches-bounces+kyle=kyleam.com@gnu.org Sender: "Guix-patches" To: Marius Bakke Cc: 29181-done@debbugs.gnu.org Hi, Marius Bakke skribis: > Ludovic Court=C3=A8s writes: > >> This reduces the nar size (as shown by 'guix size') by 124 MiB, from >> 169 MiB to 45 MiB (almost divided by 4!). > > Wow, nice catch. If I read the code correctly, it detects hard links > and replaces them with symlinks. Yes. > Could we do this unconditionally in a gnu-build-system phase? Are > there any legitimate uses of hard links in outputs? Good question. There might be situations where the software relies on having regular files (not symlinks), so I would rather do it on a case-by-case basis than have a standard phase. WDYT? Besides, a phase that looks at all the files would have to be efficient than this na=C3=AFve implementation. > That said, the patch LGTM. Pushed as dcc00f54c619118d11982383102d2e9a1b86d080, thanks! Ludo=E2=80=99.