bug#61364: core-updates branch needed multiple hacks to work

[Liliana Marie Prikler]

Am Mittwoch, dem 08.02.2023 um 09:57 +0100 schrieb Dr. Arne
Babenhauserheide:
> Hi,
> 
> I updated my Guix system to the core-updates branch because I need
> Mesa 22 and Linux 6 for my new CPU+graphics card.
Note that running core-updates is not recommended, because it is often
missing changes that hit master or staging first.  For instance, Linux
6 is on master right now (but Mesa 22 isn't).  In general, the workflow
for staging/core-updates is to first merge master into the branch, then
deal with the conflicts, then merge the branch onto master.  You can
skip the third step if you're not aiming to push the result to master.

> To get that to work required multiple hacks — mostly ignoring tests
> around missing OSErrors on sockets.
Needless to say, blindly ignoring tests is not a good idea.  Use the
without-tests transformation locally instead.  For CI on the other
hand, we should maximize the number of tests that can run reproducibly.

> I attached the patches I required in the hopes that they can help get
> the core-updates branch into the condition needed to release from it.
Please clean up your patches before submitting and try to follow the
Contribution Guidelines.  For instance, you have three patches in
sequence targeting the documentation of git.  Is it really necessary to
keep two failed attempts at fixing whatever issue this phase causes?

Cheers

bug#61329: Info

[Simon Tournier]

Hi,

On lun., 06 févr. 2023 at 14:36, Michela Francesconi 
 wrote:

> But it doesn't work. What I'm doing wrong? You can help me?

Welcome!

The purpose of the email list help-g...@gnu.org is for getting help.
Maybe it would be more suitable for asking help instead of reporting a
bug. :-)

I hope the message by Parnikkapore is answering your question.

Well, what do you think if this report is closed and instead the
discussion is happening in help-g...@gnu.org?  Since I am missing if it
is related to a bug. ;-)

Cheers,
simon

bug#56669: enhancement: Link guix system and guix home

[Andrew Tropin]

On 2022-07-26 12:23, Andrew Tropin wrote:

> On 2022-07-21 19:25, Maxime Devos wrote:
>
>> On 21-07-2022 19:13, Andrew Tropin wrote:
>>
>>> The source code is here:
>>> https://git.sr.ht/~abcdw/rde/commit/c5b4097ab99309ace23e40d957e9fa1f938f97e9
>>
>> What's the 'guix-home-gc-roots' for? I would expect the reference 
>> #$(file-append he "/activate") to be sufficient to keep things from 
>> being gc'ed.
>
> It was needed while I was testing manual activation without shepherd
> service, not needed anymore, already removed it locally.
>
>>
>>> + 
>>> 
>>>  
>>> (start #~(make-forkexec-constructor + 
>>> 
>>>  
>>> '(#$(file-append he "/activate")) + 
>>> 
>>>  
>>> #:user #$user + 
>>> 
>>>  
>>> #:environment-variables + 
>>> 
>>>  
>>> (list (string-append "HOME=" (passwd:dir (getpw #$user + 
>>> 
>>>  
>>> #:group (group:name (getgrgid (passwd:gid (getpw #$user))
>> I'm wondering if GUIX_LOCPATH is needed as well. Anyway, if not done 
>> already internally by /activate, you could consider doing it in a 
>> container to reduce potential irreproducibility, or insecurity on 
>> multi-user systems (I'd assume the #:user + #:group to be sufficient for 
>> security, especially if it appears sufficient for other system services, 
>> but I'm not some expert on what things need to be set).
>>
> It's not set by /activate.
>
>>> + 
>>> 
>>>  
>>> (provision (list (symbol-append 'guix-home- (string->symbol user + 
>>> 
>>>  
>>> (one-shot? #t) + 
>>> 
>>>  
>>> (auto-start? #f)
>> Wouldn't it then be possible for the user to login via the login manager 
>> before initialisation has completed, as gdm etc don't wait for 
>> guix-home-... currently?
>
> You are right, the same as the first one, needed for more manual
> approach, changed to #t, thank you.
>
> Three patches for this service to work is on the way on guix-patches.
> In the meantime, will try to build livecd with the home environment
> inside.
>
> P.S. Probably this system service is far from final version of this
> feature, I still think about making home-environment a part of
> user-account.  Will evaluate pros and cons, after I get livecd built
> successfully.

Sorry for the long status update, some life moments are happened.

Polished all the things on Guix Home side and I can confirm that the
service works correctly and it's possible to make home-environments a
part of operating-system record.

Current very simple implementation works relatively good.  It accepts a
list of ("user" . home-env) pairs and creates a shepherd services, which
activate respective home environments.
https://git.sr.ht/~abcdw/rde/tree/9175c7b37b6861095bae4a696aa1faadf9dc572a/src/gnu/services/home.scm#L1

This is how sway graphical environment activation is implemented in rde-live 
image.
http://files.trop.in/rde/

I still find it not completely satisfying because activation happens
when one-shot shepherd service get started and not during system
activation, which leads to the problem mentioned by Maxim: you can login
into user's shell before home-environment activated.  I would like to
just extend system activation with calls to home activation scripts, but
it's not that straightforward because we depend on user-homes (which is
a shepherd service).

That said the guix-home system service works fine and you can already
use it, but before merging it to Guix I would like to move home
activations into system activation, which requires some work on
user-homes.  It doesn't seem to be a big task, but still require some
dedication and IDK when I get spare time for it.  Let me know if this
feature blocks you in some way, otherwise I'll keep working on it in my
own pace.

-- 
Best regards,
Andrew Tropin


signature.asc
Description: PGP signature

bug#61367: guix copy fails after copying with unauthorized public key

[Jonathan Brielmaier]

Hi,

I think it would be nice if `guix copy` would check for the public key
before actually copying data.

Currently it's like this:
$ guix copy --to=user@host /gnu/store/bla
guix copy: sending 439 store items (2.984 MiB) to 'host'...
[ takes some minutes ]
guix copy: error: unauthorized public key: (public-key
 (ecc
  (curve Ed25519)
  (q #KEYXYZ#)
  )
 )

It would be nice if it would fail upfront.

~Jonathan

bug#61329: Info

[Parnikkapore via Bug reports for GNU Guix]

Hello, and welcome to Guix!

The "proper" way to run software installed using Guix is to activate the Guix 
profile; see https://guix.gnu.org/manual/en/html_node/Getting-Started.html for 
details. This sets $PATH for you, along with other environment variables that 
make sure e.g. new man pages can be found.

For setting $PATH manually, "~" is a shorthand for your home directory, so one 
of these two should work:

a)
PATH=$PATH:/home/hub_galli/mfrancesconi/.guix-profile/bin
export PATH

b)
PATH=$PATH:~/.guix-profile/bin
export PATH

Hope this helps!

bug#61364: core-updates branch needed multiple hacks to work

[Dr. Arne Babenhauserheide]

Hi,

I updated my Guix system to the core-updates branch because I need Mesa
22 and Linux 6 for my new CPU+graphics card.

To get that to work required multiple hacks — mostly ignoring tests
around missing OSErrors on sockets.

I attached the patches I required in the hopes that they can help get
the core-updates branch into the condition needed to release from it.

From 193dca1b55b68aa883c3ed8b28bf19e9527fa065 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Sat, 28 Jan 2023 22:29:39 +0100
Subject: [PATCH] gnu: tcsh: comment out test of nice with nice guix builder

* gnu/packages/patches/tcsh-fix-autotest.patch: comment out test of nice. To
  reproduce:
  guix shell tcsh -- nice -n 4 tcsh -c "nice echo 1" # works
  guix shell tcsh -- nice -n 5 tcsh -c "nice echo 1" # breaks
---
 gnu/packages/patches/tcsh-fix-autotest.patch | 29 
 1 file changed, 29 insertions(+)

diff --git a/gnu/packages/patches/tcsh-fix-autotest.patch b/gnu/packages/patches/tcsh-fix-autotest.patch
index 9f5790641b..252560dbb2 100644
--- a/gnu/packages/patches/tcsh-fix-autotest.patch
+++ b/gnu/packages/patches/tcsh-fix-autotest.patch
@@ -1,3 +1,32 @@
+--- tests/commands.at
 tests/commands.at
+@@ -890,15 +890,16 @@
+ TCSH_UNTESTED([newgrp])
+ 
+ 
+-AT_SETUP([nice])
+-
+-# Nothing really tested
+-AT_CHECK([tcsh -f -c 'nice set var=1; echo $?var'], ,
+-[0
+-])
+-
+-
+-AT_CLEANUP
++# XXX This test fails if the Guix worker has a nice value >= 5
++# AT_SETUP([nice])
++# 
++# # Nothing really tested
++# AT_CHECK([tcsh -f -c 'nice -n +1 set var=1; echo $?var'], ,
++# [0
++# ])
++# 
++# 
++# AT_CLEANUP
+ 
+ 
+ AT_SETUP([nohup])
+
 --- tests/commands.at
 +++ tests/commands.at
 @@ -921,26 +921,27 @@ AT_CLEANUP
-- 
2.39.1

From 0a1f8507a4c9a5fee9da18fcc25238289d19aa45 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Sat, 4 Feb 2023 22:30:14 +0100
Subject: [PATCH 02/15] disable glib tests

---
 gnu/packages/glib.scm | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gnu/packages/glib.scm b/gnu/packages/glib.scm
index 3d8a25be9e..31c6e73aa0 100644
--- a/gnu/packages/glib.scm
+++ b/gnu/packages/glib.scm
@@ -240,6 +240,7 @@ (define glib
"debug"))
 (arguments
  (list
+  #:tests? #f
   #:disallowed-references
   (cons tzdata-for-tests
 ;; Verify glib-mkenums, gtester, ... use the cross-compiled
-- 
2.39.1

From 45c24935a24158a68acfabcf072d81cbb4239816 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Sun, 5 Feb 2023 02:44:48 +0100
Subject: [PATCH 03/15] no tests

---
 gnu/packages/gnome.scm | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gnu/packages/gnome.scm b/gnu/packages/gnome.scm
index 0b1f30edc8..36f0e7a29b 100644
--- a/gnu/packages/gnome.scm
+++ b/gnu/packages/gnome.scm
@@ -4451,6 +4451,7 @@ (define-public vala
 (build-system glib-or-gtk-build-system)
 (arguments
  (list
+  #:tests? #f
   #:configure-flags #~(list "CC=gcc" "--enable-coverage")
   #:phases
   #~(modify-phases %standard-phases
-- 
2.39.1

From 121e2ba9ad13fe70bcf6fd7b83378d8df4b0aba1 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Sun, 5 Feb 2023 09:38:57 +0100
Subject: [PATCH 04/15] native-input docbook-utils in git

---
 gnu/packages/version-control.scm | 1 +
 1 file changed, 1 insertion(+)

diff --git a/gnu/packages/version-control.scm b/gnu/packages/version-control.scm
index 98b80b6073..86ef8860e5 100644
--- a/gnu/packages/version-control.scm
+++ b/gnu/packages/version-control.scm
@@ -255,6 +255,7 @@ (define-public git
;; For subtree documentation.
("asciidoc" ,asciidoc)
("docbook-xsl" ,docbook-xsl)
+   ("docbook-utils" ,docbook-utils)
("xmlto" ,xmlto)
("pkg-config" ,pkg-config)))
 (inputs
-- 
2.39.1

From ddc7d6dc05b75116d8f259455cf93e2bc4a027f1 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Mon, 6 Feb 2023 00:15:06 +0100
Subject: [PATCH 05/15] docbook-utils as inputs

---
 gnu/packages/version-control.scm | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/gnu/packages/version-control.scm b/gnu/packages/version-control.scm
index 86ef8860e5..f0156391dd 100644
--- a/gnu/packages/version-control.scm
+++ b/gnu/packages/version-control.scm
@@ -255,11 +255,11 @@ (define-public git
;; For subtree documentation.
("asciidoc" ,asciidoc)
("docbook-xsl" ,docbook-xsl)
-   ("docbook-utils" ,docbook-utils)
("xmlto" ,xmlto)
("pkg-config" ,pkg-config)))
 (inputs
  `(("curl" ,curl)
+   ("docbook-utils" ,docbook-utils)
("expat" ,expat)
("openssl" ,openssl)
("perl" ,perl)
-- 
2.39.1

From 4dcb3f1ba3d7ee3fdf2bf5c13e31aa0974814344 Mon Sep 17 00:00:00 2001
From: Arne Babenhauserheide 
Date: Mon, 6 Feb 2023 22:44:07 +0100
Subject: [PATCH 06/15] no docs for git

---
 gnu/packages/version-control.scm | 10 --
 1 file changed, 10 deletions(-)

diff --git a/gnu/packages/version-control.scm