From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andreas Enge Subject: Re: guix-gc Date: Thu, 10 Jan 2013 11:04:21 +0100 Message-ID: <201301101104.22164.andreas@enge.fr> References: <8738yfp4j7.fsf@gnu.org> Mime-Version: 1.0 Content-Type: multipart/alternative; boundary="Boundary-01=_lIp7QIUSYvkuXSo" Content-Transfer-Encoding: 7bit Return-path: Received: from eggs.gnu.org ([208.118.235.92]:40750) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1TtEzx-0006OI-RA for bug-guix@gnu.org; Thu, 10 Jan 2013 05:04:37 -0500 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1TtEzu-0003ZJ-QE for bug-guix@gnu.org; Thu, 10 Jan 2013 05:04:33 -0500 In-Reply-To: <8738yfp4j7.fsf@gnu.org> List-Id: Bug reports for GNU Guix List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: bug-guix-bounces+gcggb-bug-guix=m.gmane.org@gnu.org Sender: bug-guix-bounces+gcggb-bug-guix=m.gmane.org@gnu.org To: bug-guix@gnu.org --Boundary-01=_lIp7QIUSYvkuXSo Content-Type: text/plain; charset="utf-8" Content-Transfer-Encoding: quoted-printable Am Sonntag, 6. Januar 2013 schrieb Ludovic Court=C3=A8s: > Commit fe8ff02 adds a =E2=80=98guix-gc=E2=80=99 command to invoke the GC,= similar to > part of the =E2=80=98nix-store=E2=80=99 command. Below is a copy of the = manual. >=20 > Comments welcome! When running "guix-gc" without specifying any root, the complete content of= =20 /nix/store was classified as dead. This looks a bit excessive to me; should= =20 not at least the gnu build system that would be used to build a package=20 with the currently installed guix be preserved? Even if the user has not=20 installed gcc in his environement, for instance, the "current" gcc should=20 be kept, whereas any older gcc (I had quite a few of them corresponding to= =20 several versions of guix) should be deleted. If I see it correctly, this means that the output of guix-build hello --dry-run should be kept in the store. I also got the following error on the first run: deleting `/nix/store/j29s9129di8lzx7xbc9rj0ywx05pivij-readline-6.2' error: build failed: getting status of=20 `/nix/store/j29s9129di8lzx7xbc9rj0ywx05pivij-readline-6.2': No such file or= =20 directory But the next run worked and deleted everything. Andreas --Boundary-01=_lIp7QIUSYvkuXSo Content-Type: text/html; charset="utf-8" Content-Transfer-Encoding: quoted-printable

Am Sonntag,= 6. Januar 2013 schrieb Ludovic Court=C3=A8s:

> Commit= fe8ff02 adds a =E2=80=98guix-gc=E2=80=99 command to invoke the GC, similar= to

> part o= f the =E2=80=98nix-store=E2=80=99 command. Below is a copy of the manual.<= /p>

>

> Commen= ts welcome!

&nb= sp;

When runnin= g "guix-gc" without specifying any root, the complete content of = /nix/store was classified as dead. This looks a bit excessive to me; should= not at least the gnu build system that would be used to build a package wi= th the currently installed guix be preserved? Even if the user has not inst= alled gcc in his environement, for instance, the "current" gcc sh= ould be kept, whereas any older gcc (I had quite a few of them correspondin= g to several versions of guix) should be deleted.

&nb= sp;

If I see it= correctly, this means that the output of

guix-bui= ld hello --dry-run

should be k= ept in the store.

&nb= sp;

&nb= sp;

I also got = the following error on the first run:

&nb= sp;

deleting `/= nix/store/j29s9129di8lzx7xbc9rj0ywx05pivij-readline-6.2'

error: buil= d failed: getting status of `/nix/store/j29s9129di8lzx7xbc9rj0ywx05pivij-re= adline-6.2': No such file or directory

&nb= sp;

But the nex= t run worked and deleted everything.

&nb= sp;

Andreas

&nb= sp;

--Boundary-01=_lIp7QIUSYvkuXSo--