[PATCH] gnu: %bootstrap-coreutils&co: Patch egrep/fgrep to point to the correct grep.

  • Done
  • quality assurance status badge
Details
5 participants
  • Efraim Flashner
  • Jonathan Brielmaier
  • Ludovic Courtès
  • manolis837
  • zimoun
Owner
unassigned
Submitted by
manolis837
Severity
normal
Merged with
M
M
manolis837 wrote on 23 Feb 2017 12:41
(address . guix-patches@gnu.org)(address . efraim@flashner.co.il)
20170223114119.18004-1-manolis837@gmail.com
Hello everyone,

Commit 653add37af10b0d5dbc20c8bc7ce30eec8cf4ae7 patches egrep/fgrep to point to the location of grep.
But the problem is that when it was run on Hurd, the path for grep in egrep/fgrep had this form:
/gnu/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-grep-2.25/bin/grep

The above was not being substituted and caused the build to fail. So I changed the pattern to
"/[^ ]+/bin/grep" and now the path is always replaced.

Also (getcwd) returns the current path which the build takes place. This happens to be /tmp/guix-build-..-bootstrap-binaries-0.
This is wrong because it should point to the final place in /gnu/store. That why I am using "out".

Thank you,
Manolis
E
E
Efraim Flashner wrote on 25 Feb 2017 19:58
C634411A-54AB-496B-97CC-4C7D7FFF2BE2@flashner.co.il
On February 23, 2017 1:41:18 PM GMT+02:00, manolis837@gmail.com wrote:
Toggle quote (14 lines)
>
>Hello everyone,
>
>Commit 653add37af10b0d5dbc20c8bc7ce30eec8cf4ae7 patches egrep/fgrep to
>point to the location of grep.
>But the problem is that when it was run on Hurd, the path for grep in
>egrep/fgrep had this form:
>/gnu/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-grep-2.25/bin/grep
>
>The above was not being substituted and caused the build to fail. So I
>changed the pattern to
>"/[^ ]+/bin/grep" and now the path is always replaced.
>

I also pushed a patch that patched the generated [e|f]grep to not have the path inserted into it, so the aarch64 bootstrap binaries have:
#!/bin/sh
exec grep -F...
I think we're going to have to find a way that works for both architectures. I tried using (%current-target) but I couldn't get it to work inside the snippet.

Toggle quote (5 lines)
>Also (getcwd) returns the current path which the build takes place.
>This happens to be /tmp/guix-build-..-bootstrap-binaries-0.
>This is wrong because it should point to the final place in /gnu/store.
>That why I am using "out".

While that is probably better, in the end it doesn't really matter I believe, when I ran 'find /gnu -name fgrep -execdir cat {} +' it showed both bash and grep pointing to the bootstrap binaries.

Toggle quote (3 lines)
>Thank you,
>Manolis

--
Sent from my Android device with K-9 Mail. Please excuse my brevity.
L
L
Ludovic Courtès wrote on 26 Mar 2017 15:53
control message for bug #25849
(address . control@debbugs.gnu.org)
87tw6gp2kf.fsf@gnu.org
merge 25849 25848
J
J
Jonathan Brielmaier wrote on 16 Nov 2020 18:59
[PATCH] gnu: %bootstrap-coreutils
(address . 25848@debbugs.gnu.org)
62265294-3ea4-f72b-439d-0a207e5674da@web.de
@Manolis: Does this problem still exist with all the new Hurd work from
Janneke and Co?
Z
Z
zimoun wrote on 5 Jan 2022 00:57
Re: bug#25848: [PATCH] gnu: %bootstrap-coreutils&co: Patch egrep/fgrep to point to the correct grep.
(address . manolis837@gmail.com)
86o84r9hqy.fsf@gmail.com
Hi Manolis,

I am doing bug triage and I hit this one [1].



On Thu, 23 Feb 2017 at 13:41, manolis837@gmail.com wrote:

Toggle quote (10 lines)
> Commit 653add37af10b0d5dbc20c8bc7ce30eec8cf4ae7 patches egrep/fgrep to point to the location of grep.
> But the problem is that when it was run on Hurd, the path for grep in egrep/fgrep had this form:
> /gnu/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-grep-2.25/bin/grep
>
> The above was not being substituted and caused the build to fail. So I changed the pattern to
> "/[^ ]+/bin/grep" and now the path is always replaced.
>
> Also (getcwd) returns the current path which the build takes place. This happens to be /tmp/guix-build-..-bootstrap-binaries-0.
> This is wrong because it should point to the final place in /gnu/store. That why I am using "out".

Is it still relevant after the work by janneke among many others? And
exposed in these blog post [2,3].




Cheers,
simon
E
E
Efraim Flashner wrote on 5 Jan 2022 09:23
(name . zimoun)(address . zimon.toutoune@gmail.com)
YdVVdYxpBra1OHye@3900XT
On Wed, Jan 05, 2022 at 12:57:25AM +0100, zimoun wrote:
Toggle quote (27 lines)
> Hi Manolis,
>
> I am doing bug triage and I hit this one [1].
>
> 1: <http://issues.guix.gnu.org/issue/25848>
>
>
> On Thu, 23 Feb 2017 at 13:41, manolis837@gmail.com wrote:
>
> > Commit 653add37af10b0d5dbc20c8bc7ce30eec8cf4ae7 patches egrep/fgrep to point to the location of grep.
> > But the problem is that when it was run on Hurd, the path for grep in egrep/fgrep had this form:
> > /gnu/store/eeeeeeeeeeeeeeeeeeeeeeeeeeeeeeee-grep-2.25/bin/grep
> >
> > The above was not being substituted and caused the build to fail. So I changed the pattern to
> > "/[^ ]+/bin/grep" and now the path is always replaced.
> >
> > Also (getcwd) returns the current path which the build takes place. This happens to be /tmp/guix-build-..-bootstrap-binaries-0.
> > This is wrong because it should point to the final place in /gnu/store. That why I am using "out".
>
> Is it still relevant after the work by janneke among many others? And
> exposed in these blog post [2,3].
>
>
> 2: <https://guix.gnu.org/en/blog/2020/a-hello-world-virtual-machine-running-the-hurd/>
> 3: <https://guix.gnu.org/en/blog/2020/childhurds-and-substitutes/>
>

Judging that the Hurd, aarch64 and powerpc are all have Guix working on
them I assume it's working well enough that we can close this bug.

--
Efraim Flashner <efraim@flashner.co.il> ????? ?????
GPG key = A28B F40C 3E55 1372 662D 14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted
-----BEGIN PGP SIGNATURE-----
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=Qg8G
-----END PGP SIGNATURE-----


Z
Z
zimoun wrote on 5 Jan 2022 10:00
(name . Efraim Flashner)(address . efraim@flashner.co.il)
86pmp68sl5.fsf@gmail.com
Hi,

On Wed, 05 Jan 2022 at 10:23, Efraim Flashner <efraim@flashner.co.il> wrote:
Toggle quote (6 lines)
> On Wed, Jan 05, 2022 at 12:57:25AM +0100, zimoun wrote:
>> On Thu, 23 Feb 2017 at 13:41, manolis837@gmail.com wrote:

> Judging that the Hurd, aarch64 and powerpc are all have Guix working on
> them I assume it's working well enough that we can close this bug.

Therefore, I am closing.


Thanks,
simon
Closed
M
M
Manolis Ragkousis wrote on 5 Jan 2022 18:23
(name . zimoun)(address . zimon.toutoune@gmail.com)
CAFtzXzM2p=T9=H=2k=BztV4kMX=cTm4j8zfrs_2Ou3WFtVRoGQ@mail.gmail.com
Thank you, yes it can be closed.

On Wed, 5 Jan 2022, 11:11 zimoun, <zimon.toutoune@gmail.com> wrote:

Toggle quote (16 lines)
> Hi,
>
> On Wed, 05 Jan 2022 at 10:23, Efraim Flashner <efraim@flashner.co.il>
> wrote:
> > On Wed, Jan 05, 2022 at 12:57:25AM +0100, zimoun wrote:
> >> On Thu, 23 Feb 2017 at 13:41, manolis837@gmail.com wrote:
>
> > Judging that the Hurd, aarch64 and powerpc are all have Guix working on
> > them I assume it's working well enough that we can close this bug.
>
> Therefore, I am closing.
>
>
> Thanks,
> simon
>
Attachment: file
Closed
?