Github user afs commented on the pull request:
https://github.com/apache/jena/pull/110#issuecomment-167159338
These are some fairly drastic to get `ColumnMap` to work here.
`Triple` and `Quad` are central classes used everywhere in Jena so the
implications are far ranging.
Are there cost implications? I'm not sure the compiler implications of
using arrays rather than object slots. For example, there are checks on array
access - sometime they can be removed, some times not. The objects are larger
by several slots.
And it all adds many new methods to Triple and Quad. They are tuple-like
but not full tuples (`Tuple` was not designed for this).
But Tuple of 3 is not a Triple. Different `.equals`.
I'm not clear this PR meets the objective of avoiding packing and
unpacking. There is implicit creation which I don't understand the implication
of via varargs parameters. e.g. `QuadOrdering.map(quad)` unpacks and calls the
varargs method.
What is more, the naming spills over into other parts it's less clear than
before: `first` is clearer than `nodes[0]` for example. While the original is
more verbose in places, it's repetitive (which is tested extensively now) to be
replaced by creating Tuples so it is not free. [I'm not sure that I actually
understand _find anyway.]
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
