On 05/31/2018 12:10 AM, Nathaniel Smith wrote:
On Wed, May 30, 2018 at 11:14 AM, Marten van Kerkwijk
- When it comes to the core ufunc machinery, we have a limited
complexity budget. I'm nervous that if we add too many bells and
whistles, we'll end up writing ourselves into a corner where we have
trouble maintaining it, where it becomes difficult to predict how
different features interact, it becomes increasingly difficult for
third-parties to handle all the different features in their
__array_ufunc__ methods...
Re: implenetation complexity, I just want to bring up multiple-dispatch
signatures again, where the new signature syntax would just be to join
some signatures together with "|", and try them in order until one works.
I'm not convinced it's better myself, I just wanted to make sure we are
aware of it. The translation from the current proposed syntax would be:
Current Syntax Multiple-dispatch syntax
(n|1),(n|1)->() <===> (n),(n)->() | (n),()->() | (),(n)->()
(m?,n),(n,p?)->(m?,p?) <===> (m,n),(n,p)->(m,p) |
(n),(n,p)->(p) |
(m,n),(n)->(m) |
(n),(n)->()
Conceivably, multiple-dispatch could reduce code complexity because we
don't need all the special flags like UFUNC_CORE_DIM_CAN_BROADCAST, and
instead of handling special syntax for ? and | and any future syntax
separately, we just need a split("|") and then loop with the old
signature handling code.
On the other hand the m-d signatures are much less concise and the
intention is perhaps harder to read. Yet they more explicitly state
which combinations are allowed, while with '?' syntax you might have to
puzzle it out.
Cheers,
Allan
_______________________________________________
NumPy-Discussion mailing list
NumPy-Discussion@python.org
https://mail.python.org/mailman/listinfo/numpy-discussion
