On Sat, Feb 14, 2015 at 1:09 AM, Stephen J. Turnbull <stephen@xemacs.org> wrote:

Chris Barker writes:

...

merging two dicts certainly is _something_ like addition.

So is set union (merging two sets). But in Python, binary + is used mainly for - numerical addition (and element-wise numerical addition, in Counter), and - sequence concatenation. Being unordered, dicts are more like sets than sequences, so the idea of supporting dict1 + dict2 but not set1 + set2 makes me queasy. Cheers, Nathan

But you can't say what that something is with precision, even if you abstract, and nobody contests that different applications of dicts have naively *different*, and in implementation incompatible, "somethings".

So AFAICS you have to fall back on "my editor forces me to type all the characters, so let's choose the interpretation I use most often so cI can save some typing without suffering too much cognitive dissonance."

Following up the off-topic comment:

...

[In numpy,] we really want readable code:

y = a*x**2 + b*x + c

really reads well, but it does create a lot of temporaries that kill performance for large arrays. You can optimize that by hand by doing something like:

y = x**2 y *= a y += b*x y += c

Compilers can optimize such things very well, too. I would think that a generic optimization to the compiled equivalent of

try: y = x**2p y *= a y += b*x y += c except UnimplementedError: y = a*x**2 + b*x + c

would be easy to do, possibly controlled by a pragma (I know Guido doesn't like those, but perhaps an extension PEP 484 "Type Hints" could help here).

_______________________________________________ Python-ideas mailing list Python-ideas@python.org https://mail.python.org/mailman/listinfo/python-ideas Code of Conduct: http://python.org/psf/codeofconduct/