Closed ClemensLinnhoff closed 6 months ago
It is definitely strange that "optional" fields are checked as "mandatory" by the validator. I like the proposal with machine readable additions to the fields for the validator. What are your thoughts @pmai and @jdsika ?
I mean someting like "version number" and "time stamp" are somewhat mandatory?
We need an initial draft for the logic of what is optional and what is not. If radar then lidar not mandatory ? What are the edge cases? How to document this?
I will take this for the next CCB and come back with some thoughts on this.
We discussed this issue today and the CCB would encourage to start with a PR that works on the rules.yaml adding "is_set" and to start with the fields that are already marked or 100% clear to be mandatory.
We will create an example and then also adapt the mentioned section 2.2.1 in the documentation.
works on the rules.yaml adding "is_set" and to start with the fields that are already marked
It is kind of strange, that is_set is not in the rules.yml, as it is already used, e.g. in vehicle_attributes. And the osi-validation rules2yml script already parses the is_set rule somehow. I will look into it.
I created a Draft PR (see above) as a proposal. I mainly set version, timestamp and id fields as mandatory, as well as the main mounting positions.
This branch of osi-validation can handle the new input and only checks for "is_set" if it is actually a rule in the proto files.
Describe the problem
As indicated by section 2.2.1 of the OSI documentation:
The general case in the industry is however, that not all fields are set. It is completely up to the users, which fields to set in the interface.
Ask your question
Should some fields be marked as mandatory in any case? For example the version or the timestamp? If so, I propose to write this into the documentation of the respective field in machine readable form, so a checker can parse this as a rule, analog to rules like
is_greater_than_or_equal_to
.