feature CtQuery#filterChildren(null)

[pvojtechovsky]

Support for calling filterChildren without filter. It returns all visited elements.

[pvojtechovsky]

What is the problem? I do not know how to fix it.

[pvojtechovsky]

Filter filter = new ...;
element.map(new ParameterScopeQuery).map(e->filter.matches(e))...

Martin wrote: I like it, esp with lambdas, it's clear and explicit. Let's first aim at this simplest version.

There is one more problem when Filter is used in existing CtQuery#map(CtFunction). It is that query engine will throw ClassCastException, when Filter input type does not matches.

So you have to write

Filter filter = new ...;
element.map(new ParameterScopeQuery)
  .map(e->filter.matches(e))
  .failurePolicy(QueryFailurePolicy.IGNORE)...

Do you still like it? I vote for one of these methods

• CtQuery#filter(Filter)
• CtQuery#mapFilter(Filter)

Note: we cannot implement CtQuery#map(Filter), because it would avoid usage of lambdas, because of ambiguity with CtQuery#map(CtFunction)

Or any other solution - if somebody has one.

[monperrus]

super atomic PR, thanks.

Would you add one line of Javadoc in filterChildren to document the contract?

[monperrus]

It is that query engine will throw ClassCastException, when Filter input type does not matches.

which is the default behavior we want, because it's very explicit at runtime, right?

[pvojtechovsky]

It is that query engine will throw ClassCastException, when Filter input type does not matches.

which is the default behavior we want, because it's very explicit at runtime, right?

right. We want to throw CCE on standard mapping functions, because otherwise it is very hard to found "Why my query does not work and returns nothing..."

[pvojtechovsky]

Done from my point of view

[monperrus]

rebase?

[pvojtechovsky]

Done.