!DEkEPGCygOsVHrOCNQ:cadair.com

Python in Heliophysics

32 Members
A new room for collaborations in Python tools for all heliophysics http://heliopython.org4 Servers

Load older messages


Timestamp Message
26 Nov 2018
14:44:18@ehsteve:matrix.orgehsteve Btw I’m pretty close to summarizing my unconference sessions. If you participate in either of them please have a look at my report for accuracy. Thanks.
29 Nov 2018
16:59:27@jtniehof:matrix.orgjtniehofI have my unconference summaries up, just focused on quick "what we talked about" and the recs. Thanks to Dan and Baptiste for note-taking. Further checks to see if I captured things appropriately are appreciated.
30 Nov 2018
17:50:38@jtniehof:matrix.orgjtniehofI did some googling and any Python 4 references are pie-in-the-sky guesses. (Also for reference it's been 10 years since python 3 came out and 7 years since numpy had Python 3 support.)
7 Dec 2018
21:16:40@deadzero:matrix.orgsamaloney joined the room.
10 Dec 2018
16:53:56@cadair:cadair.comCadair If anyone is around AGU but not currently in AGU then I am hunting a coffee shop in the general area of the conference centre
12 Dec 2018
20:39:14@jtniehof:matrix.orgjtniehofDrat. We were recommending .data as exposing the fundamental data type (e.g. if you use a numpy array internally it should expose the numpy array) but that doesn't work for is-a data types. numpy already uses .data to expose the internal buffer, so something that is-a numpy array can't redefine that to return a (vanilla) numpy array without breaking the ability to act like a numpy array itself.
20:41:56@jtniehof:matrix.orgjtniehofhttps://docs.scipy.org/doc/numpy/reference/generated/numpy.ndarray.data.html
20:48:48@jtniehof:matrix.orgjtniehofMaybe change that to .data for composited objects and something else for subclassed...I'd suggest .base but numpy already uses that (to refer to the array that owns this array's memory.)
20:49:08@jtniehof:matrix.orgjtniehof.super?
20:49:26@cadair:cadair.comCadair I am not entirely sure you need a .data on an subclassed array
20:49:31@cadair:cadair.comCadairor similar
20:49:58@jtniehof:matrix.orgjtniehofIt's purely syntactic sugar; it's so if people are used to doing .attrs/.meta for metadata and .data for data they get something reasonable.
20:52:42@jtniehof:matrix.orgjtniehofand it can be nice to have something that gets you back to the base class (although I think it should be possible to make asarray do it reasonably without a copy, need to look into that.)
20:53:13@cadair:cadair.comCadairwhat are you subclassing ndarray for out of interest?
20:53:52@jtniehof:matrix.orgjtniehofso we can add attributes/metadata
20:54:17@jtniehof:matrix.orgjtniehof(and then we add all the magic to make sure they get pickled properly and such, too)
20:54:21@cadair:cadair.comCadairWhat library?
20:54:27@jtniehof:matrix.orgjtniehofspacepy
20:54:34@cadair:cadair.comCadairsubclassing ndarray is a special kind of hell fwiw
20:54:45@jtniehof:matrix.orgjtniehofyou don't have to tell me twice.....well aware of that
20:54:46@cadair:cadair.comCadairas it seems you know / are learning
20:54:48@cadair:cadair.comCadairlol
20:55:10@cadair:cadair.comCadairI am not sure what to suggest
20:56:18@cadair:cadair.comCadairstandarding between encapsulated and subclassed is an interesting problem
20:56:18@jtniehof:matrix.orgjtniehoffor now I think the solution is to wash some dishes and see what comes to me
20:56:29@cadair:cadair.comCadairalways a good option!
1 Jan 2019
14:46:45@cadair:cadair.comCadair Reminder: apply for Python in Astronomy 2019 before January 31st: https://t.co/PPwnyR3ie3 . More info at https://t.co/8dajKOgfPI
11 Jan 2019
17:10:11@wtbarnes:matrix.orgwtbarnesIs there a telecon right now?
17:10:42@cadair:cadair.comCadair🤷
17:14:29@wtbarnes:matrix.orgwtbarnesI guess it will remain a mystery then

There are no newer messages yet.


Back to Room List