Auto save pipely

[MaxKarthan]

At the moment pipely does not auto save new configurations.
After some evaluations its clear that the save button has to be removed and replaced with a autosave function.

For the autosave function it is necessary to properly distinguish pipeline drafts from running pipeline instances and handle mulitple versions of a configuration.

We need a concept for the visual representation of multiple instances and versions of a single configuration.

This issue is related to #54

[kKdH]

I’m a bit confused. Why is the auto-save feature related to the feature of running multiple instances of a pipeline?

Does auto-save mean automatic deployment of a pipeline?

[MaxKarthan]

I’m a bit confused. Why is the auto-save feature related to the feature of running multiple instances of a pipeline?

Actually... it's not xD
My bad, we do not necessarily need to be able run multiple instances with the same config to change the underlying configuration of a running pipeline. We just need versioning of the configurations.

Does auto-save mean automatic deployment of a pipeline?

No. I just thought about saving the configuration every time a change appears. But then we need to save a new version of the configuration on every change or maybe one new version per session and overwrite this new version on every change during this single session.
But with this behavior a user would select a running instance change its configuration but this wouldn't affect the actual configuration of the underlying and running pipeline. But would create a new configuration version. This does not seem to be self-explaining.

I think we need a clear separation between instance and configuration, maybe editing a instance just should not be possible. So you could only edit configurations. Then we could create two different overviews. One with all the running pipelines where it is possible to monitor the state etc., change the version of the pipelines configuration and so on and one with all configurations where one can edit a configuration and create new versions of existing ones.

This would of course bring some major disadvantages for example the live view on the data would not be possible from within the editor if you have no connection to a running instance etc.

At the moment I have no proper solution but It's something we should think about. Maybe it's more a subject for a personal meeting than a discussion on GitHub^