issue_comments

Table actions

3 rows where issue = 911393744 sorted by updated_at descending

✎ View and edit SQL

This data as json, CSV (advanced)

Suggested facets: created_at (date), updated_at (date)

user 3

author_association 2

issue 1

  • Broadcast does not return Datasets with unified chunks · 3 ✖
id html_url issue_url node_id user created_at updated_at ▲ author_association body reactions performed_via_github_app issue
862578251 https://github.com/pydata/xarray/issues/5435#issuecomment-862578251 https://api.github.com/repos/pydata/xarray/issues/5435 MDEyOklzc3VlQ29tbWVudDg2MjU3ODI1MQ== dcherian 2448579 2021-06-16T17:41:12Z 2021-06-16T17:41:12Z MEMBER

Can we outsource the chunking decision to dask somehow?

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   Broadcast does not return Datasets with unified chunks 911393744
855375205 https://github.com/pydata/xarray/issues/5435#issuecomment-855375205 https://api.github.com/repos/pydata/xarray/issues/5435 MDEyOklzc3VlQ29tbWVudDg1NTM3NTIwNQ== malmans2 22245117 2021-06-06T10:25:07Z 2021-06-06T10:25:07Z CONTRIBUTOR

So under the hood use dask broadcast_to(..., chunks=None) and users should run unify_chunks before and/or after broadcasting?

In the example above, if my target is chunksize (1, 1), ds.unify_chunks() works. Does it make any difference passing chunks=(1, 1) to dask broadcast_to rather than using chunks=None and then rechunk?

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   Broadcast does not return Datasets with unified chunks 911393744
855346745 https://github.com/pydata/xarray/issues/5435#issuecomment-855346745 https://api.github.com/repos/pydata/xarray/issues/5435 MDEyOklzc3VlQ29tbWVudDg1NTM0Njc0NQ== shoyer 1217238 2021-06-06T06:25:01Z 2021-06-06T06:25:01Z MEMBER

Thanks for the report!

I'm not entirely sure Xarray should always make chunks that same when broadcasting. In some cases this could make a lot of very small chunks, which could lead to a slow-down.

One alternative (discussed in https://github.com/pydata/xarray/issues/3371) would be to make a top-level unify_chunks function, that would unify chunks in multiple Dataset and/or DataArray object, independent of broadcasting/alignment.

 
{
    "total_count": 0,
    "+1": 0,
    "-1": 0,
    "laugh": 0,
    "hooray": 0,
    "confused": 0,
    "heart": 0,
    "rocket": 0,
    "eyes": 0
}
   Broadcast does not return Datasets with unified chunks 911393744

Advanced export

JSON shape: default, array, newline-delimited, object

CSV options: 

CREATE TABLE [issue_comments] (
   [html_url] TEXT,
   [issue_url] TEXT,
   [id] INTEGER PRIMARY KEY,
   [node_id] TEXT,
   [user] INTEGER REFERENCES [users]([id]),
   [created_at] TEXT,
   [updated_at] TEXT,
   [author_association] TEXT,
   [body] TEXT,
   [reactions] TEXT,
   [performed_via_github_app] TEXT,
   [issue] INTEGER REFERENCES [issues]([id])
);
CREATE INDEX [idx_issue_comments_issue]
    ON [issue_comments] ([issue]);
CREATE INDEX [idx_issue_comments_user]
    ON [issue_comments] ([user]);