reduce(func, seq, initial=0)
Hi folk, I stumbled upon the builtin reduce function. It has an optional parameter initial, however if trying to set it like import operator as op reduce(op.add, range(10), initial=0) I get TypeError: reduce() takes no keyword arguments I think it would be really straightforward to also add keyword possibilities, at least to the parameter initial. What do you think? best, Stephan
On Fri, Oct 10, 2014 at 8:11 PM, Stephan Sahm <Stephan.Sahm@gmx.de> wrote:
The first thing to note is that reduce(), in Python 3, is now imported from functools; if you're working with Python 2, you won't see this changed unless you can convince people that this is a bug to be fixed. However, prefixing your code with "from functools import reduce" produces the exact same result in Python 3. The question is, though: What's the advantage? The first two arguments are mandatory, and there's only one optional argument. Do you need to say "initial=" on that? ChrisA
On 10 Oct 2014 20:07, "Chris Angelico" <rosuav@gmail.com> wrote:
On Fri, Oct 10, 2014 at 8:11 PM, Stephan Sahm <Stephan.Sahm@gmx.de> wrote:
I stumbled upon the builtin reduce function. It has an optional
parameter
Converting functools to argument clinic for 3.5 should fix it. I don't believe anyone has tackled that yet. Cheers, Nick.
On Fri, Oct 10, 2014 at 11:11:36AM +0200, Stephan Sahm wrote:
Unfortunately, this is a limitation of many built-in functions and methods, possibly the majority. E.g.: Help on built-in function getattr in module builtins: getattr(...) getattr(object, name[, default]) -> value but: py> getattr(None, name='spam', default=42) Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: getattr() takes no keyword arguments I seem to recall that there was a reason for this, but I don't quite remember what... searching the archives may give a clue. Personally, I have no objections *in principle* to adding support for keyword arguments to built-ins, but it will depend on the details of why they weren't already supported. -- Steven
* Steven D'Aprano <steve@pearwood.info> [2014-10-10 21:10:10 +1100]:
FWIW, I'd also really like to see keyword arguments for these builtins. I always have to keep myself from doing things like "foo bar baz".split(count=1) But as said, I suspect there is *some* reason this is not possible. Florian -- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG 0xFD55A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/
* Stefan Behnel <stefan_ml@behnel.de> [2014-10-10 13:29:51 +0200]:
Oh, thanks for the heads-up! The name was from memory, and I guess the last time I tested this was with 3.3. Florian -- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG 0xFD55A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/
On 10 Oct 2014 21:07, "Florian Bruhin" <me@the-compiler.org> wrote:
The technical reason is that many older CPython functions written in C use PyArg_ParseTuple, which only supports positional arguments. When there's a concrete readability gain from adding keyword argument support to a particular function, patches will often be accepted. Switching to argument clinic also offers improved introspection support, so that can be a good way to pursue the enhancement. Regards, Nick.
On Fri, Oct 10, 2014 at 8:11 PM, Stephan Sahm <Stephan.Sahm@gmx.de> wrote:
The first thing to note is that reduce(), in Python 3, is now imported from functools; if you're working with Python 2, you won't see this changed unless you can convince people that this is a bug to be fixed. However, prefixing your code with "from functools import reduce" produces the exact same result in Python 3. The question is, though: What's the advantage? The first two arguments are mandatory, and there's only one optional argument. Do you need to say "initial=" on that? ChrisA
On 10 Oct 2014 20:07, "Chris Angelico" <rosuav@gmail.com> wrote:
On Fri, Oct 10, 2014 at 8:11 PM, Stephan Sahm <Stephan.Sahm@gmx.de> wrote:
I stumbled upon the builtin reduce function. It has an optional
parameter
Converting functools to argument clinic for 3.5 should fix it. I don't believe anyone has tackled that yet. Cheers, Nick.
On Fri, Oct 10, 2014 at 11:11:36AM +0200, Stephan Sahm wrote:
Unfortunately, this is a limitation of many built-in functions and methods, possibly the majority. E.g.: Help on built-in function getattr in module builtins: getattr(...) getattr(object, name[, default]) -> value but: py> getattr(None, name='spam', default=42) Traceback (most recent call last): File "<stdin>", line 1, in <module> TypeError: getattr() takes no keyword arguments I seem to recall that there was a reason for this, but I don't quite remember what... searching the archives may give a clue. Personally, I have no objections *in principle* to adding support for keyword arguments to built-ins, but it will depend on the details of why they weren't already supported. -- Steven
* Steven D'Aprano <steve@pearwood.info> [2014-10-10 21:10:10 +1100]:
FWIW, I'd also really like to see keyword arguments for these builtins. I always have to keep myself from doing things like "foo bar baz".split(count=1) But as said, I suspect there is *some* reason this is not possible. Florian -- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG 0xFD55A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/
* Stefan Behnel <stefan_ml@behnel.de> [2014-10-10 13:29:51 +0200]:
Oh, thanks for the heads-up! The name was from memory, and I guess the last time I tested this was with 3.3. Florian -- http://www.the-compiler.org | me@the-compiler.org (Mail/XMPP) GPG 0xFD55A072 | http://the-compiler.org/pubkey.asc I love long mails! | http://email.is-not-s.ms/
On 10 Oct 2014 21:07, "Florian Bruhin" <me@the-compiler.org> wrote:
The technical reason is that many older CPython functions written in C use PyArg_ParseTuple, which only supports positional arguments. When there's a concrete readability gain from adding keyword argument support to a particular function, patches will often be accepted. Switching to argument clinic also offers improved introspection support, so that can be a good way to pursue the enhancement. Regards, Nick.
participants (6)
-
Chris Angelico -
Florian Bruhin -
Nick Coghlan -
Stefan Behnel -
Stephan Sahm -
Steven D'Aprano