ungleich-gitea

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:24 +00:00

__service: Suggestions on improving the type

mentioned in merge request !847

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:23 +00:00

__service: Suggestions on improving the type

You're be right with "a single type for all implementations" which makes the usage more intuitive. But there also exists the opposite: __package_* types as example. It's a bigger mess to have many…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:21 +00:00

__service: Suggestions on improving the type

any real world example why one should need or use __service?

I can only think one usecase, but this isn't supported by cdist.

__some_type foo --doing something

onchange='__some_type/foo'…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:20 +00:00

__service: Suggestions on improving the type

Difference between the different __package_* and __service_* types I can see are: The different package implementations

have quite a bit of code,
take different arguments.

While for service…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:19 +00:00

__service: Suggestions on improving the type

I can see three real-world use cases:

you install a package and want to make sure that the service is running after installation,
you need to stop a service to uninstall a package,
you changed…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:17 +00:00

__service: Suggestions on improving the type

I require to restart/reload a service. For my case, I generate a dns configuration and reload the dns daemon, so the changes will be applied. Also referring to the manual, the chapter "Drive into…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:16 +00:00

__service: Suggestions on improving the type

@matze You raise some valid points, but they all show shortcomings of cdist, don't they?

__service being a type is unfortunate for types which do their changes in code-remote and not through…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:15 +00:00

__service: Suggestions on improving the type

My suggestion is how I made the __systemd_service type (see !844). States are sure, because this is implemented by cdist's design.

The thing you referred from puppet looks like hooks which will…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:13 +00:00

__service: Suggestions on improving the type

I've looked around why there's a special __config_file type, but the reason is to reload a daemon configurations if the content of the current file changes. This is done by passing a command, which…

ungleich-gitea commented on issue ungleich-public/cdist#62

2021-11-20 15:33:12 +00:00

__service: Suggestions on improving the type

mentioned in issue #843

ungleich-gitea commented on issue ungleich-public/cdist#60

2021-11-20 15:33:10 +00:00

[logging]: DEBUG: ${__target_host}: return_output is True, ignoring stdout

rationale: side effect
desirable to mute: yes
support for creating mr to support scripts that don't output stdout: yes

ungleich-gitea commented on issue ungleich-public/cdist#60

2021-11-20 15:33:09 +00:00

[logging]: DEBUG: ${__target_host}: return_output is True, ignoring stdout

Just checked the code and I am a bit confused: we have return_output and depending on PIPE. Best thing is though that I can use git blame to blame @poljakowski for it: