Add option to item-directive to hide specific relation-types

[SteinHeselmans]

Add option to item-directive to hide specific relation-types in the rendered item definition.

[Letme]

Can you also add a test for your new feature?

Should be something short, where no-show would not have link relationship there, while without it by default it would (be asserted). Maybe noshow could be better name as hiderelationship or hidelinks ? What is your use case?

[SteinHeselmans]

I'll check if i can add a test for it, and consider another name. Not a fan of 'hiderelationship' as you might want to reuse it for hiding attributes as well in the future.

Use case: I have documentation items for every release, and for every feature that we implement. Those link to eachother to mark which features go into which release. We also have documentation items for the V-model stuff (requirements, design, test, ...). The V-model-items link to the feature-item that impacts (added/altered) the feature implementation. So that way, we have some sort of history in our V-model doc. Now, you can imagine that the feature-items can have a lot of links as they link to everything in the V-model documentation. To make those links easier to interpret, i added an item-matrix for every layer in the V-model: feature to requirement, feature to design, feature to test, ... These tables give more sorted output and is more organized compared to what we see on the item rendering itself. So i want to hide (or notshow ;-) ) the relations to the V-model on the epic, to remove this clutter.

[Letme]

If I am looking over the docs then we have a lot of hide stuff, so :hidesource: ,hidetarget, :hidetitle: in item-matrixes so maybe better would be :hidelinks: ?

Looking deeper, we also do have :nocaptions: and :nooverwrite: for example (yay, no consistency) so we could have :nolinks: or noattributes. You do not want to mix attributes and relationships in one flag, so I would keep in mind we want hideattributes option if there would be a use case for this.

In traceability we have sourcetype so :hidetype: (or :notype: since items have no - dont like this one)?

[SteinHeselmans]

I think hidelinks is most consistent amongst what you found, and indeed hideattributes might be a good addition if we ever need it. The hardest part is done, we found a name ;-)

Still working on a new test...

[JasperCraeghs]

Hi Stein! I hope you are well. Thank you for contributing.

We never use the term "link" in the name of an option. I am strongly in favor of hidetype.

$ grep -F 'type:' -r mlx -I
mlx/traceability/directives/item_relink_directive.py:        if not reverse_type:
mlx/traceability/directives/item_relink_directive.py:         :type: relationship_type
mlx/traceability/directives/item_link_directive.py:         :type: relationship_type
mlx/traceability/directives/item_tree_directive.py:         :type: <<relationship>> ...
mlx/traceability/directives/item_2d_matrix_directive.py:         :type: <<relationship>> ...
mlx/traceability/directives/item_pie_chart_directive.py:         :sourcetype: <<relationship>> ...
mlx/traceability/directives/item_pie_chart_directive.py:         :targettype: <<relationship>> ...
mlx/traceability/directives/item_pie_chart_directive.py:         :splitsourcetype:
mlx/traceability/directives/item_matrix_directive.py:         :type: <<relationship>> ...
mlx/traceability/directives/item_matrix_directive.py:         :sourcetype: <<relationship>> ...

[SteinHeselmans]

i think all comments are handled, i also added a test. this became an end-to-end test running sphinx completely, as i didn't want to be bothered implementing a test cycling through the docutils objects...

[Letme]

Love the whole sphinx path test. Thanks for your contribution and for improving our test coverage 😉

I will let Jasper do another review as he had more comments, but LGTM.