No specific error raised when features required not available in plugin

[PinkShnack]

When trying to access a plugin feature in a dataset that does not contain a features required feature, the only error raised is "Feature 'FEATURE_NAME' does not exist!" from dclab.rtdc_dataset.core.__getitem__. This is currently expected, but would it be reasonable to have a more specific exception?

[paulmueller]

A verbose error message would have to name all the ancillary features that implement the feature and a message that states the user should check whether all conditions for that feature are met (e.g. print req_required, config_required, etc.). There are also custom functions a plugin feature may implement.

[PinkShnack]

There are also custom functions a plugin feature may implement.

By this do you mean the "method check required" or some function that for example just checks that the correct features are available in the rtdc_ds?

If you mean the latter, when I tried, the plugin didn't raise the error I expected. I will recheck this and if the problem persists then I will paste the copy here...

[paulmueller]

Yes, I am referring to this:

https://github.com/ZELLMECHANIK-DRESDEN/dclab/blob/0.37.3/dclab/rtdc_dataset/ancillaries/ancillary_feature.py#L71-L83

[PinkShnack]

Perhaps we should add an example of this for plugin features in the docs and tests?