ARROW-12706: [Python] Drop Python 3.6 support#12026
ARROW-12706: [Python] Drop Python 3.6 support#12026pitrou wants to merge 4 commits intoapache:masterfrom
Conversation
|
@github-actions crossbow submit py |
|
@github-actions crossbow submit -g wheel |
|
|
|
Revision: 636cd9c Submitted crossbow builds: ursacomputing/crossbow @ actions-1336
|
|
Revision: 636cd9c Submitted crossbow builds: ursacomputing/crossbow @ actions-1336
|
|
@github-actions crossbow submit py |
|
Revision: b0664f0 Submitted crossbow builds: ursacomputing/crossbow @ actions-1338
|
|
@github-actions crossbow submit -g wheel |
|
Revision: b0664f0 Submitted crossbow builds: ursacomputing/crossbow @ actions-1339
|
|
@github-actions crossbow submit test-ubuntu-20.04-python-3 |
|
Revision: df03b26 Submitted crossbow builds: ursacomputing/crossbow @ actions-1340
|
|
@github-actions crossbow submit -g wheel |
|
Revision: e6a2ea8 Submitted crossbow builds: ursacomputing/crossbow @ actions-1341
|
| import collections | ||
| ordered_dict = collections.OrderedDict | ||
| # Starting with Python 3.7, dicts are guaranteed to be insertion-ordered. | ||
| ordered_dict = dict |
There was a problem hiding this comment.
Not necessarily here, but I suppose we can also remove ordered_dict alltogether, and use dict(..) instead in the few places it is being used?
There was a problem hiding this comment.
ordered_dict has a richer API than dict, so theoretically we cannot return dict instead without breaking compatibility. Not sure that matters in practice, though.
There was a problem hiding this comment.
ordered_dicthas a richer API thandict, so theoretically we cannot returndictinstead without breaking compatibility. Not sure that matters in practice, though.
Ah, yes indeed. It's currently being used in Table/RecordBatch.to_pydict and KeyValueMetadata.to_dict.
Personally, I would maybe just break this at some point
|
Benchmark runs are scheduled for baseline = c00767c and contender = 0f4d662. 0f4d662 is a master commit associated with this PR. Results will be available as each benchmark for each run completes. |
No description provided.