RTree creation API might seem a bit contraintuitive about rtree HOT 3 CLOSED

Thanks for your input and I'm glad you are finding the library useful.

So the static method approach means you can't use method chaining so code like:

tree = RTree.create()
                   .add(1, point(2,3))
                   .add(2, point(5,2))
                   .add(3, point(9,4));

becomes much more verbose and less discoverable to make happen with static add methods:

tree = RTreeBuilder.add(RTreeBuilder.create(), 1, point(2,3));
tree = RTreeBuilder.add(tree, 2, point(5,2));
tree = RTreeBuilder.add(tree, 3, point(9,4));

For an immutable data structure I would never expect add to do anything but return a new object otherwise it's not immutable and you make that point too. add in this case means create a new RTree based on the current tree with the item added to it. Perhaps a better name for the method or I could emphasise what add is doing in the documentation in the README. I don't like the idea of using a builder because it just makes for unnecessary typing and I'm not keen to lose method chaining.

from rtree.

Actually I thought about a normal builder, so you could chain and use it with build() in the end like .add().add().build(). But then I realized that you often want to perform online modifications so a builder is not really suited for that... And the static method for that purpose might be also not so nice, I agree.

Just emphasizing what the immutable operations do in the README would probably be fine then.

from rtree.

Righto, I'll update the README. Thanks for your input.

On 13 March 2015 at 10:05, Pere Ferrera [email protected] wrote:

Actually I thought about a normal builder, so you could chain and use it
with build() in the end like .add().add().build(). But then I realized that
you often want to perform online modifications so a builder is not really
suited for that... And the static method for that purpose might be also not
so nice, I agree.

Just emphasizing what the immutable operations do in the README would
probably be fine then.

—
Reply to this email directly or view it on GitHub
#20 (comment).

from rtree.