From mboxrd@z Thu Jan 1 00:00:00 1970 From: YOANN P Subject: RE: Questions regarding "Relocatable" option Date: Thu, 21 Jun 2018 18:12:51 +0000 Message-ID: References: , <87woutp2im.fsf@gnu.org> Mime-Version: 1.0 Content-Type: text/plain; charset="Windows-1252" Content-Transfer-Encoding: quoted-printable Return-path: Received: from eggs.gnu.org ([2001:4830:134:3::10]:33317) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1fW44r-0000TV-2a for guix-devel@gnu.org; Thu, 21 Jun 2018 14:13:02 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1fW44p-00024e-9V for guix-devel@gnu.org; Thu, 21 Jun 2018 14:13:01 -0400 In-Reply-To: <87woutp2im.fsf@gnu.org> Content-Language: fr-FR 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: =?Windows-1252?Q?Ludovic_Court=E8s?= Cc: "guix-devel@gnu.org" Hi ludo,=0A= =0A= >I experimented with it a bit a reported my findings here:=0A= >=0A= >=A0 https://lists.gnu.org/archive/html/guix-devel/2018-05/msg00139.html=0A= >=0A= >In short it=92s still inconvenient, so it won=92t happen for this release.= =0A= >=0A= >To address the main limitation, I thought we could have a=0A= >=93--relocatable=94 package transformation option as well=B9 that people c= ould=0A= >use to automatically wrap what they install.=A0 Food for thought=85=0A= >=0A= >=B9 https://www.gnu.org/software/guix/manual/html_node/Package-Transformat= ion-Options.html=0A= =0A= Sounds great even if it is inconvenient for now :) sadly, not see many repl= ies to your test :/=0A= =0A= >We already discussed this a while back, I think.=A0 :-) I think the=0A= >default should remain unchanged given the extra overhead (in space and=0A= >build time, not just startup time of the resulting binaries) that=0A= >wrappers introduce, and given that user namespaces are missing on many=0A= >systems still.=0A= =0A= indeed, wasn't aware it was you though ^^ but I don't think we talked about= space and build time overhead when we had this discussion.=0A= Namespace will be activate everywhere with (a lot of) time and eventually w= ith the introduction of proot as fallback wrapper :p=0A= So, the best option (only from my point of view), it will be to have 2 bina= ry build downloadable : =0A= - normal=0A= - relocatable=0A= So every users ( rootless or not ) could use guix as an environnement/packa= ges manager with the overhead involved depending on the binary downloaded.= =0A= With having the choice of the build, the user could choose overhead or not = depending the env he's on, and could deploy his "manifest" he had previousl= y put in scm ( oh yeahhhh ) to deploy his environment.=0A= =0A= Thanks for all of your hard work on this project Ludo=0A= =0A= Best regards,=0A= Yoann=0A= =0A= De : Ludovic Court=E8s =0A= Envoy=E9 : mercredi 20 juin 2018 21:32=0A= =C0 : YOANN P=0A= Cc=A0: guix-devel@gnu.org=0A= Objet : Re: Questions regarding "Relocatable" option=0A= =A0=0A= Hello Yoann,=0A= =0A= YOANN P skribis:=0A= =0A= > - Could we hope to see it included in the next release ?=0A= =0A= It=92s definitely gonna be there.=A0 :-)=A0 Guix is mostly rolling release,= in fact.=0A= =0A= > - Could we hope to see it included by default in the binary tar.gz of thi= s next release to be able to use guix directly in an unprivileged environme= nt ? ( i dreaming of this every night ^^ )=0A= =0A= I experimented with it a bit a reported my findings here:=0A= =0A= =A0 https://lists.gnu.org/archive/html/guix-devel/2018-05/msg00139.html=0A= =0A= In short it=92s still inconvenient, so it won=92t happen for this release.= =0A= =0A= To address the main limitation, I thought we could have a=0A= =93--relocatable=94 package transformation option as well=B9 that people co= uld=0A= use to automatically wrap what they install.=A0 Food for thought=85=0A= =0A= =B9 https://www.gnu.org/software/guix/manual/html_node/Package-Transformati= on-Options.html=0A= =0A= > - Does the "relocatable" could be the default option and have an "--no-re= locatable" option for people who want to avoid the little extra time to sta= rt an application ?=0A= =0A= We already discussed this a while back, I think.=A0 :-) I think the=0A= default should remain unchanged given the extra overhead (in space and=0A= build time, not just startup time of the resulting binaries) that=0A= wrappers introduce, and given that user namespaces are missing on many=0A= systems still.=0A= =0A= Thanks,=0A= Ludo=92.=0A=