**Breaking**: data_kind: data is None and required now returns the 'empty' kind #3482
Conversation
…data types fall back to 'vectors'
pygmt/clib/session.py
Outdated
| "image": tempfile_from_image, | ||
| "stringio": self.virtualfile_from_stringio, | ||
| "matrix": self.virtualfile_from_matrix, | ||
| "none": self.virtualfile_from_vectors, |
There was a problem hiding this comment.
Can we change to another word like "empty" to reduce confusion with None?
There was a problem hiding this comment.
I have another idea to completely eliminate the "none" kind. I'll open a POC PR later.
There was a problem hiding this comment.
I have another idea to completely eliminate the "none" kind. I'll open a POC PR later.
It turns out the idea can't eliminate the "none" kind. So let's focus on this PR again.
Can we change to another word like "empty" to reduce confusion with
None?
I want to emphasize that data_kind(data=None) returns "none", so kind == "none" means data is None, which IMHO is very intuitive. That's why I choose "none" as the new kind name, but I'm also OK with a new kind name if others think "empty" is a better name.
There was a problem hiding this comment.
The intention of kind == "none" is to indicate that the data is stored in the x, y, z variables rather than in data (i.e. data is None means data is in xyz variables). I thought about naming it as kind="xyz" but since we are planning to support other 1d arrays like intensity, symbol, etc (#1076), maybe we should call it kind="xyzarrays"? Or I'm ok with using kind="empty" too, just don't want to use "none" because it is confusing with None.
There was a problem hiding this comment.
maybe we should call it
kind="xyzarrays"
Although not documented, I think in our codes, when we talk about "arrays", we usually mean numpy arrays, so a better name is "xyzvectors", but I feel it's confusing with the "vectors" kind. So, rename none to empty in a1e67d3.
seisman
changed the title
weiji14
added
final review call
Co-authored-by: Wei Ji <[email protected]>
seisman
removed
the
final review call
…odule performance (#3502) Co-authored-by: Wei Ji <[email protected]>
…das<=2.1 (#3505) Co-authored-by: Wei Ji <[email protected]>
…to_matrix/vectors_to_arrays/array_to_datetime (#3496)
Co-authored-by: Wei Ji <[email protected]>
…a_kind/vectors-none
seisman
force-pushed
the
data_kind/vectors-none
branch
from
80c5aa5 to
a469acf
Compare
Need to wait for #3351 or merge directly into #3351.
This PR adds a new kind
"empty"for thedata=None, required=Truecase. The new kind name makes more sense and can simplify the logic of codes.