[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Tobias Geerinckx-Rice via Guix-patches via]

* gnu/installer/services.scm (%system-services): Add CUPS.
---
 gnu/installer/services.scm | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)

diff --git a/gnu/installer/services.scm b/gnu/installer/services.scm
index b0007218ae..195b27aa77 100644
--- a/gnu/installer/services.scm
+++ b/gnu/installer/services.scm
@@ -2,6 +2,7 @@
 ;;; Copyright © 2018 Mathieu Othacehe <m.othacehe@gmail.com>
 ;;; Copyright © 2019 Ludovic Courtès <ludo@gnu.org>
 ;;; Copyright © 2020 Jan (janneke) Nieuwenhuizen <janneke@gnu.org>
+;;; Copyright © 2021 Tobias Geerinckx-Rice <me@tobias.gr>
 ;;;
 ;;; This file is part of GNU Guix.
 ;;;
@@ -117,7 +118,13 @@
      (system-service
       (name (G_ "DHCP client (dynamic IP address assignment)"))
       (type 'network-management)
-      (snippet '((service dhcp-client-service-type)))))))
+      (snippet '((service dhcp-client-service-type))))
+
+     ;; Dealing with documents.
+     (system-service
+      (name (G_ "CUPS printing system (Web interface disabled)"))
+      (type 'document)
+      (snippet '((service cups-service-type)))))))
 
 (define (desktop-system-service? service)
   "Return true if SERVICE is a desktop environment service."
-- 
2.32.0

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Mathieu Othacehe]

Hey Tobias,

> +      (name (G_ "CUPS printing system (Web interface disabled)"))
> +      (type 'document)
> +      (snippet '((service cups-service-type)))))))

Any particular reason not to enable the Web interface by default?

Otherwise, if you made sure that 

--8<---------------cut here---------------start------------->8---
make check-system TESTS="gui-installed-os gui-uefi-installed-os
gui-installed-os-encrypted gui-installed-desktop-os-encrypted"
--8<---------------cut here---------------end--------------->8---

works, I think you can go ahead.

Thanks,

Mathieu

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Tobias Geerinckx-Rice via Guix-patches via]

[-- Attachment #1: Type: text/plain, Size: 1245 bytes --]

Hi Mathieu,

Mathieu Othacehe 写道：
> Hey Tobias,
>
>> +      (name (G_ "CUPS printing system (Web interface 
>> disabled)"))
>> +      (type 'document)
>> +      (snippet '((service cups-service-type)))))))
>
> Any particular reason not to enable the Web interface by 
> default?

I think it unwise to override upstream defaults for a feature they 
consider ‘on life support’.

Their position is that instead of the bolted-on CGI interface, 
users should use a UI that speaks directly to CUPS's C or HTTP 
API, like system-config-printer.

The only reason I added the default state in brackets is to remind 
those who do want the Web UI that it's a separate thing, so when 
they open https://localhost:631 a week later they might remember 
that and what to enable :-)

> Otherwise, if you made sure that 
>
> make check-system TESTS="gui-installed-os gui-uefi-installed-os
> gui-installed-os-encrypted gui-installed-desktop-os-encrypted"

Eh, I just ran all of the tests.  No new ones failed.

By the way: the *os-encrypted ones never work, they all hang 
forever at the Linux LUKS prompt.

> works, I think you can go ahead.

Thanks for replying!  I'll push it tomorrow.

Kind regards,

T G-R

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 247 bytes --]

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Mathieu Othacehe]

Hey Tobias,

>> Otherwise, if you made sure that 
>> make check-system TESTS="gui-installed-os gui-uefi-installed-os
>> gui-installed-os-encrypted gui-installed-desktop-os-encrypted"
>
> Eh, I just ran all of the tests.  No new ones failed.

On Berlin, all four installer tests are now failing. The log message
seems to point towards the recent CUPS service addition:
https://ci.guix.gnu.org/build/639140/log/raw.

Mathieu

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Mathieu Othacehe]

> On Berlin, all four installer tests are now failing. The log message
> seems to point towards the recent CUPS service addition:
> https://ci.guix.gnu.org/build/639140/log/raw.

There are a few pointers on how to fix the test failure in this thread:
https://issues.guix.gnu.org/47979#8.

Thanks,

Mathieu

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Tobias Geerinckx-Rice via Guix-patches via]

[-- Attachment #1: Type: text/plain, Size: 1380 bytes --]

Mathieu,

Thanks.  I finally pushed [0] to fix this.  Finally, because I 
really wanted to filter other services with (const #t) instead of 
#f, but didn't find a way to make it work.  I mean, …

  (define %extra-packages
    (cons* cups
           …
           ((@@ (gnu services cpus) cups-configuration-extensions)
            (cups-configuration))))

…just, no.

I'm sure my sub-par Gexp-foo is to blame, and there's a clean way 
to include the closure of a service in the installation-os…?

Mathieu Othacehe 写道：
> On Berlin, all four installer tests are now failing. The log 
> message
> seems to point towards the recent CUPS service addition:
> https://ci.guix.gnu.org/build/639140/log/raw.

All tests ‘passed‘ (in retrospect obviously not seeing the cups 
changes) on my laptop, until I rebuilt my checkout from scratch.

Recompiling Guix as a matter of course is prohibitive on this 
machine, but I'll make a habit of doing so before running the rare 
and slower system test.

> There are a few pointers on how to fix the test failure in this 
> thread:

Oops.  Oh well.  I'm actually glad I missed that.  I finally feel 
some familiarity with how system and installer tests work, now.)

Kind regards,

T G-R

[0]: 
https://git.savannah.gnu.org/cgit/guix.git/commit/?id=0dbd2c3b089d50bdfbed86916d06cc4f0da8893d

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 247 bytes --]

[bug#49201] [PATCH] installer: Offer the CUPS printing service.

[Mathieu Othacehe]

Hey Tobias,

> I'm sure my sub-par Gexp-foo is to blame, and there's a clean way to include
> the closure of a service in the installation-os…?

Including a service closure in the installation tests is currently quite
tricky. As explained here: https://issues.guix.gnu.org/47979#15, I guess
it would require to add the cups-service-type to the
installation-target-os-for-gui-tests procedure.

> Oops.  Oh well.  I'm actually glad I missed that.  I finally feel some
> familiarity with how system and installer tests work, now.)

The CI also confirms that things are back to normal.

Thanks for taking care of that,

Mathieu