[
https://issues.apache.org/jira/browse/ARROW-1614?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17219199#comment-17219199
]
Rok Mihevc edited comment on ARROW-1614 at 10/22/20, 6:20 PM:
--------------------------------------------------------------
As proposed by [~jorisvandenbossche] I've made a draft PR (
https://github.com/apache/arrow/pull/8510) with python logic prototype. It was
heavily inspired by [~bryanc]'s text-extensions-for-pandas. Once we agree on
the design we can rewrite it to c++.
As this is for the case where all tensors in the array are of the same shape I
propose we store the data in a single Tensor. Is there a good reason not to do
that?
I assume we should support non-contiguous tensors. I'll add that.
Any comments at this point?
[~chrish42] - feel free to jump in any time.
was (Author: rokm):
As proposed by [~jorisvandenbossche] I've made a draft PR (
[https://github.com/apache/arrow/pull/8510.]) with python logic prototype. It
was heavily inspired by [~bryanc]'s text-extensions-for-pandas. Once we agree
on the design we can rewrite it to c++.
As this is for the case where all tensors in the array are of the same shape I
propose we store the data in a single Tensor. Is there a good reason not to do
that?
I assume we should support non-contiguous tensors. I'll add that.
Any comments at this point?
[~chrish42] - feel free to jump in any time.
> [C++] Add a Tensor logical value type with constant dimensions, implemented
> using ExtensionType
> -----------------------------------------------------------------------------------------------
>
> Key: ARROW-1614
> URL: https://issues.apache.org/jira/browse/ARROW-1614
> Project: Apache Arrow
> Issue Type: New Feature
> Components: C++, Format
> Reporter: Wes McKinney
> Priority: Major
> Labels: pull-request-available
> Time Spent: 20m
> Remaining Estimate: 0h
>
> In an Arrow table, we would like to add support for a column that has values
> cells each containing a tensor value, with all tensors having the same
> dimensions. These would be stored as a binary value, plus some metadata to
> store type and shape/strides.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)