Required oneof

[dmdashenkov]

In this PR we add a new validation option: (is_required) for oneof group fields.

Previously, to make a oneof "required" (i.e. to enforce that one field is always set), the users had to write a (required_field) expression:

message Meal {
    option (required_field) = "fish | meat | veggies";

    oneof choice {
        Fish fish = 1;
        Meat meat = 2;
        Veggitables veggies = 3;
    }
}

This approach is error-prone, as the field names may be misspelled, the fields may be added or deleted, etc.

Now, the users can mark the oneof group with a (is_required) option:

message Meal {
    oneof choice {
        option (is_required) = true;

        Fish fish = 1;
        Meat meat = 2;
        Veggitables veggies = 3;
    }
}

The second snippet is equivalent to the first one but has less room for accidental errors.

Fixes #521.

[codecov]

Codecov Report

Merging #523 into master will increase coverage by 0.25%.
The diff coverage is 100.00%.

@@             Coverage Diff              @@
##             master     #523      +/-   ##
============================================
+ Coverage     73.90%   74.16%   +0.25%     
- Complexity     2912     2935      +23     
============================================
  Files           500      502       +2     
  Lines         11747    11787      +40     
  Branches        658      661       +3     
============================================
+ Hits           8682     8742      +60     
+ Misses         2841     2823      -18     
+ Partials        224      222       -2     

[dmdashenkov]

@armiol, @alexander-yevsyukov, PTAL.

[alexander-yevsyukov]

Why do we change the package?

[alexander-yevsyukov]

Currently the type package contains only few classes that deal with known types.

[alexander-yevsyukov]

There's only one issue related to the changed package. Please see my question.

[alexander-yevsyukov]

The same question regarding the package as for the class under the test. Why do change the package?

[dmdashenkov]

We move OneofDeclaration from another module. I've changed the package to a more appropriate io.spine.code.proto, which is also the package where the FieldDeclaration resides.

[dmdashenkov]

@alexander-yevsyukov, PTAL again.

[alexander-yevsyukov]

LGTM