On Sun, Oct 25, 2009 at 12:42 AM, Christopher Burns wrote:

Cool, thanks. Mind if I update the HOWTO_DOCUMENT adding in the partial list below?

Sure, that would be useful. While you're at it, could you get rid of the {True, False}?

Cheers, Ralf

Chris

On Sat, Oct 24, 2009 at 3:19 PM, Ralf Gommers wrote:

...

Not all of them are listed in one place. For general advice, see the Parameters section of http://projects.scipy.org/numpy/wiki/CodingStyleGuidelines One error on that page is {True, False}, that should be bool.

str is correct, string is not. Reason: str is the type.

Partial list: str bool list tuple sequence ndarray array_like

or if you can be more precise: list of str sequence of ints

and for keywords, add ", optional"

_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion

On Sun, Oct 25, 2009 at 1:11 AM, Christopher Burns wrote:

Done.

That section looks much better now. Except for the word "back-tics" :)

Thanks, Ralf

On Sat, Oct 24, 2009 at 3:52 PM, Ralf Gommers wrote:

...

Sure, that would be useful. While you're at it, could you get rid of the {True, False}?

---

NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion

One other comment (sorry I'm late chiming in): in general, for something like "sequence of ints," usually what is really intended as viable input is "array-like of int-likes," and indeed, in the process of confirming this for various functions, I have found bugs where what was intended was in fact not supported. So, though it's more work, i.e., will take more time, the ideal scenario, IMO, when you're dealing w/ something like that, is to confirm that the function does indeed presently support the full gamut of viable inputs, note any strange behavior, post to the list if you're uncertain if it's a bug, or just file a bug ticket if you are sure. And in the past, when this has come up, I've been instructed to document the intended behavior, not the present buggy behavior (which just reinforces the need to file a bug report). DG On Sat, Oct 24, 2009 at 4:30 PM, Christopher Burns wrote:

Just committed a change to 'backticks'.

;)

On Sat, Oct 24, 2009 at 4:17 PM, Ralf Gommers wrote:

...

That section looks much better now. Except for the word "back-tics" :)

Thanks, Ralf

---

NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion