On 27/07/16 03:17, ellisonanne wrote:
> As a very small (kind of nitty) style item, I would much rather have
> setter methods for QuerySchema to set the `elementName` and
> `filterNamesSet` rather than setting these fields via
>
> ``` querySchema.getElementNames().addAll(elementNames);
> querySchema.getFilteredElementNames().addAll(filterNamesSet); ```
>
> In the original code, these were passed in as part of the constructor
> (which has now been removed in this refactor).
Sure, I can loop back and change that.
So the thinking was that rather than parse all the XML and build the
schema in a 'big bang', the code now creates an empty schema and then
populates it incrementally as it is being parsed. I agree it is a style
thing, but IMHO incremental composition leads to less, simpler code.
I can change the parser to remember all the element names and set them
at once if you prefer.
Either way the creation of a new schema is being controlled by builders
in the o.a.pirk.schema.{data|query} packages, with a goal that they are
immutable once created [1]. So any setters would remain default
(package) scope visibility.
Do you agree that schemas should be immutable (by schema API users)?
[1] still not quite true yet today.
Regards,
Tim
