This follows on from the changes in 4985a42724.
* gnu/services/base.scm (%default-authorized-guix-keys): Add
bordeaux.guix.gnu.org.pub.
Signed-off-by: Christopher Baines <mail@cbaines.net>
Fixes <https://issues.guix.gnu.org/48521>.
The problem was caused by the 'file-system-mapping' record not being in scope.
* gnu/services/networking.scm (opendht-shepherd-service): Import the (gnu
system file-systems) module.
[requirement]: Depend on networking, to avoid spurious output.
[modules]: New field.
[start] <group>: New argument.
* gnu/services/networking.scm (maybe-number?, maybe-string?): New procedures.
(<opendht-configuration>): New configuration record.
(%opendht-accounts): New variable.
(opendht-configuration->command-line-arguments): New procedure.
(opendht-shepherd-service, opendht-service-type): New variables.
* doc/guix.texi (Networking Services): Document the new service.
This is added for convenience and for uniformity with the
define-maybe/no-serialization syntax introduced in the previous commit.
* gnu/services/configuration.scm
(define-configuration/no-serialization): New syntax.
Before this change, using define-maybe along define-configuration with the
no-serialization syntactic keyword would result in the following warning:
warning: possibly unbound variable `VARIABLE-NAME'
This change introduces the define-maybe/no-serialization variant that does
away with defining a serialization helper procedure, which makes it possible
to avoid the above warning.
* gnu/services/configuration.scm (define-maybe/no-serialization): New syntax.
(define-maybe-helper): New procedure.
(define-maybe): Define syntax using the above procedure.
* tests/services/configuration.scm (tests): Fix module name.
(custom-number-serializer): Do not print to standard output.
(maybe-number?, serialize-maybe-number): New procedures defined via the
define-maybe macro.
(config-with-maybe-number): New configuration.
(serialize-number): New procedure.
("maybe value serialization"): New test.
(maybe-string?): New procedure defined via the define-maybe/no-serialization
macro.
(config-with-maybe-string/no-serialization): New configuration.
("maybe value without serialization no procedure bound"): New test.
The non-hygienic binding of the source location accessor was set to
'-location'; in modules where multiple configurations were defined, it would
cause compilation warnings such as:
gnu/services/mail.scm:175:0: warning: shadows previous definition of
`%-location-procedure' at gnu/services/mail.scm:165:0
* gnu/services/configuration.scm (define-configuration-helper) <id>: Add a
missing #'stem argument to use the configuration name as a prefix to the
location accessor identifier.
In some cases, rather than globally disabling serialization, it may be more
appropriate to disable or otherwise alter the serialization procedure of a
specific field. In large module, multiple configurations may also exist that
would need to alter the default serialization procedure, which is named after
the field type. Being able to specify a per-field serialization procedure
provides more flexibility.
* gnu/services/configuration.scm (define-configuration): Add an optional
pattern variable to allow specifying a custom serialization procedure.
(define-configuration-helper) <field-serializer>: Use it to transform the
syntax.
(empty-serializer): New procedure.
(serialize-package): Alias to ‘empty-serializer’.
Signed-off-by: Maxim Cournoyer <maxim.cournoyer@gmail.com>
Serialization is not always useful, for example when deriving command line
arguments from a configuration. This change provides a way to turn it off,
which removes the need to define a bunch of dummy serialization procedures.
Credit goes to Andrew Gierth (RhodiumToad) from #guile for providing the
solution. Thank you!
* gnu/services/configuration.scm (define-configuration-helper): New procedure.
(define-configuration) <no-serialization>: New syntactic keyword. Use it in a
new pattern. Refactor the macro so that it makes use of the above helper
procedure.
This resolves a compilation warning introduced with commit bb716e8d9d.
* gnu/services/configuration.scm (configuration->documentation): Use display
to print the string instead of format.
The original (undocumented) procedure to generate the doc has a difficult to
grasp interface; add a simpler one on top of it.
* gnu/services/configuration.scm (configuration->documentation): New procedure.
Not all fields in a configuration have a sensible default value. This changes
makes it possible to omit a default value for a configuration field, requiring
the user to provide a value.
* gnu/services/configuration.scm (configuration-missing-field): New procedure.
(define-configuration): Make default value optional.
Signed-off-by: Maxim Cournoyer <maxim.cournoyer@gmail.com>
This fixes the following issue where spice-vdagent would fail to start if the
spice-vdagent-sock socket file already existed:
spice-vdagentd: Fatal could not create the server socket
/run/spice-vdagentd/spice-vdagent-sock: Error binding to address: Address
already in use
The requirement is also modified to depend on dbus-system, a cue taken from
upstream's own systemd service file (see 'data/spice-vdagentd.service' in the
sources).
* gnu/services/spice.scm (spice-vdagent-activation): Delete procedure.
(spice-vdagent-shepherd-service): Fix indentation.
[requirement]: Replace udev by dbus-system.
[start]: Ensure the spice-vdagentd run-time directory exists and that the
spice-vdagent-sock socket file does *not* exist before forking the daemon.
* gnu/services/mail.scm (protocol-configuration): Add an ‘imap-metadata?’
setting to enable IMAP METADATA support in the ‘imap’ protocol.
* doc/guix.texi (Mail Services): Document it.
Signed-off-by: Tobias Geerinckx-Rice <me@tobias.gr>
* gnu/services/spice.scm (spice-vdagent-activation): Update runtime directory
from /var/run/spice-vdagentd to /run/spice-vdagentd.
(spice-vdagent-service-type): Specify a default value and fix indentation.
This service doesn't create a PID file, but it does write a file with the
processed commits in it, so create a directory to contain that.
* gnu/services/guix.scm (guix-build-coordinator-queue-builds-activation):
Ensure /var/cache/guix-build-coordinator-queue-builds exists.
* gnu/services/databases.scm (mysql-configuration): Add extra-environment
(mysql-service): Use #:log-file and #:environment-variables
* doc/guix.texi: Document it.
Signed-off-by: Leo Prikler <leo.prikler@student.tugraz.at>
Use the default client port rather than the default agent communication port
for the queue builds script.
* gnu/services/guix.scm (<guix-build-coordinator-queue-builds-configuration>):
Change coordinator default.
The queue builds script doesn't create a pid file, so don't have the shepherd
expect to find one.
* gnu/services/guix.scm (guix-build-coordinator-queue-builds-shepherd-services):
Remove #:pid-file.
Don't hardcode usernames, as these are in the config. Also fix the %user being
missing from the queue-builds service activation.
* gnu/services/guix.scm (guix-build-coordinator-activation,
guix-build-coordinator-agent-activation): Use config for the user name.
(guix-build-coordinator-queue-builds-activation): Define %user.
To the agent configuration.
* gnu/services/guix.scm (<guix-build-coordinator-agent-configuration>): New
field, max-1min-load-average.
(guix-build-coordinator-agent-configuration-max-1min-load-average): New
procedure.
(guix-build-coordinator-agent-shepherd-services): If set, include the
max-1min-load-average in the agent arguments.
* gnu/services/vpn.scm (<wireguard-peer>): Add 'keep-alive' field.
(wireguard-configuration-file): Use it.
* doc/guix.texi (VPN Services): Document it.
This is a follow-up of c311147bd1.
* gnu/services/databases.scm (<postgresql-role-configuration>)[host]: Set to
"/tmp" which the default Postgresql socket directory.
* gnu/services/databases.scm (mysql-configuration): Add extra-environment
(mysql-service): Use #:log-file and #:environment-variables
* doc/guix.texi: Document it.
Signed-off-by: Leo Prikler <leo.prikler@student.tugraz.at>
Fixes <https://bugs.gnu.org/46737>.
PostgreSQL running with a different socket directory to the default one in the
package itself breaks some services, this commit restores the previous
behaviour where PostgreSQL by default will run with a socket directory that
matches the default used by PostgreSQL packaged for Guix.
Switching to a different default value can happen, but only alongside changing
the PostgreSQL package.
* gnu/services/databases.scm (<postgresql-config-file>)[socket-directory]:
Change default to #false.
* doc/guix.texi (Database Services): Update documentation, and specify a
different value for disabling connections via sockets.
* gnu/tests/guix.scm (%guix-data-service-os): Use default PostgreSQL
behaviour.
* gnu/tests/monitoring.scm (%zabbix-os): Likewise.
* gnu/tests/web.scm (patchwork-os): Likewise.
Signed-off-by: Leo Famulari <leo@famulari.name>
That way, the default config with (advertises? #t) and without a cache
will offer zstd-compressed substitutes, which should lead to much higher
throughput than gzip.
* gnu/services/base.scm (default-compression): When
'guix-publish-configuration-cache' returns true, use higher level
compression ratios. Add "zstd".
* doc/guix.texi (Base Services): Mention zstd.
