#20255 'search-paths' should respect both user and system profile.

Message #62 received at 20255@debbugs.gnu.org (full text, mbox, reply):

Received: (at 20255) by debbugs.gnu.org; 22 Nov 2015 18:44:06 +0000
From debbugs-submit-bounces@debbugs.gnu.org Sun Nov 22 13:44:06 2015
Received: from localhost ([127.0.0.1]:48512 helo=debbugs.gnu.org)
	by debbugs.gnu.org with esmtp (Exim 4.80)
	(envelope-from <debbugs-submit-bounces@debbugs.gnu.org>)
	id 1a0ZcM-0007p5-4r
	for submit@debbugs.gnu.org; Sun, 22 Nov 2015 13:44:06 -0500
Received: from mail-lf0-f43.google.com ([209.85.215.43]:35511)
 by debbugs.gnu.org with esmtp (Exim 4.80)
 (envelope-from <alezost@gmail.com>) id 1a0ZcJ-0007ow-H1
 for 20255@debbugs.gnu.org; Sun, 22 Nov 2015 13:44:04 -0500
Received: by lfdl133 with SMTP id l133so14554063lfd.2
 for <20255@debbugs.gnu.org>; Sun, 22 Nov 2015 10:44:02 -0800 (PST)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113;
 h=from:to:cc:subject:references:date:in-reply-to:message-id
 :user-agent:mime-version:content-type:content-transfer-encoding;
 bh=eFGiOv9zK8gJykCXkbE2YszzgkyUu/bxQSgKjPHDn08=;
 b=QpTegTzWVRb3NnuTCC/lnOWGwdECMmRsJJpIdYICaNPg9rfVYu5QFlEqYDaR88t8z0
 t+aJzfAeknVHVtNA0Xjc+49bW6GvA7CWVhizOBKteudtG+HnxFeW6sGnGAOxopfLmfec
 NCSJSlGeyP4LMhTkGBev3yvexPZmm4Z4XDie7ewNh55tHVZ+ZYH3slzV0S/UZnMajOF0
 i8fbou8JCF1zE1qhLUFPFpr2CKtbPm4YkPa2epaaKtE12RHDTl2gAgjIWvt1pA9DXGGv
 Mt93A+EZiobpAZ5WhfTS2m8VH4YujXkcW4lyotG5Hzt6qtvDWRX1MFL2NcaxAUv8byy7
 iuqw==
X-Received: by 10.25.21.70 with SMTP id l67mr7759908lfi.27.1448217842519;
 Sun, 22 Nov 2015 10:44:02 -0800 (PST)
Received: from leviafan ([217.107.192.146])
 by smtp.gmail.com with ESMTPSA id l81sm1148934lfb.40.2015.11.22.10.44.01
 (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128);
 Sun, 22 Nov 2015 10:44:01 -0800 (PST)
From: Alex Kost <alezost@gmail.com>
To: ludo@gnu.org (Ludovic Courtès)
Subject: Re: bug#20255: 'search-paths' should respect both user and system
 profile.
References: <877ftschjt.fsf@gmail.com> <87fv8fip01.fsf@gnu.org>
 <87d23j1bxk.fsf@gmail.com> <871tjyfnl8.fsf@gnu.org>
 <876199q4z1.fsf@gmail.com> <87ioca4ojo.fsf@gnu.org>
 <87lh9tvcws.fsf@gnu.org> <87h9kguwc4.fsf@gmail.com>
 <87ziy7d90z.fsf@gnu.org> <874mgfkxee.fsf@gmail.com>
 <87wptb5d1y.fsf@gnu.org> <87r3jisc76.fsf@gmail.com>
 <87lh9q1f2i.fsf@gnu.org>
Date: Sun, 22 Nov 2015 21:44:00 +0300
In-Reply-To: <87lh9q1f2i.fsf@gnu.org> ("Ludovic
 \=\?utf-8\?Q\?Court\=C3\=A8s\=22'\?\=
 \=\?utf-8\?Q\?s\?\= message of "Sun, 22 Nov 2015 11:52:37 +0100")
Message-ID: <877fl9q3gv.fsf@gmail.com>
User-Agent: Gnus/5.13 (Gnus v5.13) Emacs/24.5 (gnu/linux)
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: quoted-printable
X-Spam-Score: -0.7 (/)
X-Debbugs-Envelope-To: 20255
Cc: 20255@debbugs.gnu.org
X-BeenThere: debbugs-submit@debbugs.gnu.org
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: <debbugs-submit.debbugs.gnu.org>
List-Unsubscribe: <https://debbugs.gnu.org/cgi-bin/mailman/options/debbugs-submit>, 
 <mailto:debbugs-submit-request@debbugs.gnu.org?subject=unsubscribe>
List-Archive: <https://debbugs.gnu.org/cgi-bin/mailman/private/debbugs-submit/>
List-Post: <mailto:debbugs-submit@debbugs.gnu.org>
List-Help: <mailto:debbugs-submit-request@debbugs.gnu.org?subject=help>
List-Subscribe: <https://debbugs.gnu.org/cgi-bin/mailman/listinfo/debbugs-submit>, 
 <mailto:debbugs-submit-request@debbugs.gnu.org?subject=subscribe>
Errors-To: debbugs-submit-bounces@debbugs.gnu.org
Sender: "Debbugs-submit" <debbugs-submit-bounces@debbugs.gnu.org>
X-Spam-Score: -0.7 (/)

Ludovic Courtès (2015-11-22 13:52 +0300) wrote:

> Alex Kost <alezost@gmail.com> skribis:
>
>>>> At first, because of the slowdown: it may be a few hundred milliseconds
>>>> for you, but it's several seconds for me.
>>>
>>> Really?  Can you show the output of:
>>>
>>>   time guix package -p /run/current-system/profile \
>>>                     -p ~/.guix-profile --search-paths
>>
>> real	0m2.634s
>> user	0m0.568s
>> sys	0m0.080s
>
> Ouch, that’s a problem.  This suggests that this is 2 seconds of I/O.
> I’m not sure what can be done to improve that.
>
>>> In the discussion of this bug, we tried hard to avoid resorting to
>>> invoking a program, but ultimately no other solution came out.
>>
>> I don't need a solution for this bug, I just want to have an option to
>> avoid invoking "guix package --search-paths" in my "/etc/profile".
>
> Are you denying that this is a bug?  Are you denying that there’s a
> usability issue at hand?

I agree it's a usability issue.

> To me, what 宋文武 reported at the beginning of this thread is a
> usability issue.  We’ve hacked around it so far, but we know there are
> cases where the hacks aren’t enough.
>
> We could declare it as “won’t fix”, but I’m not comfortable with that.

No, no, I'm against “won't fix”.  I don't mind if it's called a bug, and
a solution you suggest is the best, but it suits only the default case
of a single user profile.  If I have several user profiles, it does
nothing useful for me, only wastes the time.

>>>> For example, when I do:
>>>>
>>>>   GUIX_IGNORE_SYSTEM_PROFILE_ENV=1 guix system build my-config.scm
>>>>
>>>> the "etc/profile" of the built system will not contain those 'eval ...'
>>>> lines.  WDYT?
>>>
>>> This would be unreasonable.  We’re talking about a basic feature here.
>>> If basic features are broken to the point that we prefer to offer ways
>>> to bypass them, and have a semi-broken system, then there’s a problem,
>>> IMO.
>>
>> Sorry, but I would really like to bypass this feature
>
> [...]
>
> I very well understand your concern, so thanks for chiming in.
> Please let’s also consider the bug at hand.

OK, for the bug at hand, invoking "guix package --search-paths" looks
like the only possible solution, but please don't commit this patch
without giving a user a chance to decide what to put in /etc/profile.

> The solution I came up with might be inadequate.  Then we need to come
> up with an alternate proposal, or to resign and mark it as “wontfix.”

It is adequate and I'm not against it.

> What would you suggest?

After all, I realized what is my main concern: "/etc/profile" is
non-editable.  If I don't like some pieces of this file, I can do
nothing, and I just have to live with it and suffer.  Ideally I would
like to decide what pieces I want to put in /etc/profile and what I
don't.  But it's probably not possible, so…

… what I suggest now is just to give an option to avoid generating the
default /etc/profile.  What about making an 'operating-system' field for
this file (similar to 'sudoers-file' or 'hosts-file')?  So when such
'profile-file' is specified, it will be used instead of the default one
(of course, it should be mentioned in the manual that it's only for
those users who are sure what they do).

If this 'profile-file' field appears, I will gladly use it, and I will
not object to any future changes in /etc/profile.

-- 
Alex

Send a report that this bug log contains spam.