id,node_id,number,title,user,state,locked,assignee,milestone,comments,created_at,updated_at,closed_at,author_association,pull_request,body,repo,type,active_lock_reason,performed_via_github_app,reactions,draft,state_reason
309471814,MDU6SXNzdWUzMDk0NzE4MTQ=,189,Ability to sort (and paginate) by column,9599,closed,0,9599,,31,2018-03-28T18:04:51Z,2018-04-15T18:54:22Z,2018-04-09T05:16:02Z,OWNER,,"As requested in https://github.com/simonw/datasette/issues/185#issuecomment-376614973

I've previously avoided this for performance reasons: sort-by-column on a column without an index is likely to perform badly for hundreds of thousands of rows.

That's not a good enough reason to avoid the feature entirely though. A few options:

* Allow sort-by-column by default, give users the option to disable it for specific tables/columns
* Disallow sort-by-column by default, give users option (probably in `metadata.json`) to enable it for specific tables/columns
* Automatically detect if a column either has an index on it OR a table has less than X rows in it

We already have the mechanism in place to cut off SQL queries that take more than X seconds, so if someone DOES try to sort by a column that's too expensive it won't actually hurt anything - but it would be nice to not show people a ""sort"" option which is guaranteed to throw a timeout error.

The vast majority of datasette usage that I've seen so far is on smaller datasets where the performance penalties of sort-by-column are extremely unlikely to show up.

----

Still left to do:

- [x] UI that shows which sort order is currently being applied (in HTML and in JSON)
- [x] UI for applying a sort order (with rel=nofollow to avoid Google crawling it)
- [x] Sort column names should be escaped correctly in generated SQL
- [x] Validation that the selected sort order is a valid column
- [x] Throw error if user attempts to apply _sort AND _sort_desc at the same time
- [x] Ability to disable sorting (or sort only for specific columns) in metadata.json
- [x] Fix ""201 rows where sorted by sortable_with_nulls "" bug
",107914493,issue,,,"{""url"": ""https://api.github.com/repos/simonw/datasette/issues/189/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,completed
312312125,MDU6SXNzdWUzMTIzMTIxMjU=,194,Rename table_rows and filtered_table_rows to have _count suffix,9599,closed,0,,,2,2018-04-08T14:53:37Z,2018-04-09T05:25:22Z,2018-04-09T05:25:22Z,OWNER,,"These fields represent counts of items:

    ""table_rows"": 131,
    ""filtered_table_rows"": 8,

But the names make it sound like they might be arrays full of rows. Adding a `_count` suffix would make this more clear:

    ""table_rows_count"": 131,
    ""filtered_table_rows_count"": 8,
",107914493,issue,,,"{""url"": ""https://api.github.com/repos/simonw/datasette/issues/194/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",,completed
312355154,MDExOlB1bGxSZXF1ZXN0MTgwMTg4Mzk3,196,_sort= and _sort_desc= parameters to table view,9599,closed,0,,,0,2018-04-09T00:07:21Z,2018-04-09T05:10:29Z,2018-04-09T05:10:23Z,OWNER,simonw/datasette/pulls/196,See #189 ,107914493,pull,,,"{""url"": ""https://api.github.com/repos/simonw/datasette/issues/196/reactions"", ""total_count"": 0, ""+1"": 0, ""-1"": 0, ""laugh"": 0, ""hooray"": 0, ""confused"": 0, ""heart"": 0, ""rocket"": 0, ""eyes"": 0}",0,