On Thu, Mar 20, 2014 at 4:55 PM, Sankarshan Mudkavi <smudk...@uwaterloo.ca>wrote:
> Yes 2) is indeed what I was suggesting. My apologies for being unclear, I > was unsure of how much detail and technical information I should include in > the proposal. > well, you need to put enough in that it's clear what it means. I think examples are critical -- at least that's how I learn things. > I'm not sure how much of a hit the performance would take if we were to > take of the Z handler. Do you have any major concerns as of now regarding > that, or do you want to wait till I provide more specific details? > more detail would be good. My comment about performance is that if numpy needs to call a Python object to do the time zone handling for each value in an array, that is going to pretty slow -- but maybe better than not having it at all. And there shouldn't be any reason not to have a fast path for when the array is naive or you are working with two arrays that are in the same TZ -- the really common case that we care about performance for. So ot probably comes down to one extra field... It also looks like the last option you mentioned seems quite reasonable > too. To only do what ISO 8601 does. Perhaps, it would be better to > implement that first and then look for an improvement later on? Do you have > a preference for this or the option 2) ? > I'm liking that one: It seems pretty easy to allow a tag for TZ offset, and not much extra math when converting. And this could be pretty useful. But I'm not writing the code... -Chris -- Christopher Barker, Ph.D. Oceanographer Emergency Response Division NOAA/NOS/OR&R (206) 526-6959 voice 7600 Sand Point Way NE (206) 526-6329 fax Seattle, WA 98115 (206) 526-6317 main reception chris.bar...@noaa.gov
_______________________________________________ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion