schemagen: nullable fields

[estr0gen]

Currently, schemagen will generate structs that do not use pointers for fields that are nullable. This results in ambiguity as you cannot tell the difference between null and false or 0.

[dkropachev]

@estr0gen , can you please provide a code example.

[estr0gen]

Sorry for the late reply, I must have missed this somehow. Basically, when you run schemagen, it should output structs that use pointers for fields that are able to be nulled on the database side. Example:

type CommentsStruct struct {
	Id             string
	Content        string
}

should generate as

type CommentsStruct struct {
  	Id             string
	Content       *string
}

assuming the Id field is the primary key and Content can be nulled. I hope this makes sense as I am not too familiar with the terminology.

[dkropachev]

Sorry for the late reply, I must have missed this somehow. Basically, when you run schemagen, it should output structs that use pointers for fields that are able to be nulled on the database side. Example:

type CommentsStruct struct {
Id string
Content string
}
should generate as

type CommentsStruct struct {
Id string
Content *string
}
assuming the Id field is the primary key and Content can be nulled. I hope this makes sense as I am not too familiar with the terminology.

How would you define what is nullable what is not, in CQL almost everything is nullable?

[estr0gen]

That is correct. I think everything that is nullable in the database should be nullable in the generated code. This means that the only not nullable fields would be primary keys and partition keys (and maybe secondary indexes(?), never used them before).