issue_comments

Also https://github.com/simonw/datasette-copyable

I released 0.49a0 which means I can update the main branches of those two plugins - I'll push a release of them once 0.49 is fully shipped.

That worked! https://github.com/simonw/sqlite-utils/runs/1043640785?check_suite_focus=true

I'm going to try doing this as a GitHub Actions test matrix.

Thanks @simonwiles, this is now released in 2.16.1: https://sqlite-utils.readthedocs.io/en/stable/changelog.html

I'm going to start running black exclusively in the GitHub Actions workflow, rather than having it run by the unit tests themselves.

Yeah I think that failure is actually because there's a brand new release of Black out and it subtly changes some of the formatting rules. I'll merge this and then run Black against the entire codebase.

Looks like it failed the "black" formatting test - possibly because there's a new release if black out. I'm going to merge despite that failure.

Thanks very much!

Maybe use transform_table() in #114 for this? Would be less efficient as it would copy the whole table but it would reduce library complexity a bit.

Thanks! And yeah, I had updating the docs on my list too :) Will try to get to it this afternoon (budgeting time is fraught with uncertainty at the moment!).

That pull request should update this section of the documentation too:

If you have more than one record to insert, the insert_all() method is a much more efficient way of inserting them. Just like insert() it will automatically detect the columns that should be created, but it will inspect the first batch of 100 items to help decide what those column types should be.

https://github.com/simonw/sqlite-utils/blob/ea87c2b943fdd162c42a900ac0aea5ecc2f4b9d9/docs/python-api.rst#L393

Here's a suggested test update: ```diff diff --git a/sqlite_utils/db.py b/sqlite_utils/db.py index a8791c3..12fa2f2 100644 --- a/sqlite_utils/db.py +++ b/sqlite_utils/db.py @@ -1074,6 +1074,13 @@ class Table(Queryable): all_columns = list(sorted(all_columns)) if hash_id: all_columns.insert(0, hash_id) + else: + all_columns += [ + column + for record in chunk + for column in record + if column not in all_columns + ] validate_column_names(all_columns) first = False # values is the list of insert data that is passed to the diff --git a/tests/test_create.py b/tests/test_create.py index a84eb8d..3a7fafc 100644 --- a/tests/test_create.py +++ b/tests/test_create.py @@ -707,13 +707,15 @@ def test_insert_thousands_using_generator(fresh_db): assert 10000 == fresh_db["test"].count

-def test_insert_thousands_ignores_extra_columns_after_first_100(fresh_db): +def test_insert_thousands_adds_extra_columns_after_first_100(fresh_db): + # https://github.com/simonw/sqlite-utils/issues/139 fresh_db["test"].insert_all( [{"i": i, "word": "word_{}".format(i)} for i in range(100)] - + [{"i": 101, "extra": "This extra column should cause an exception"}] + + [{"i": 101, "extra": "Should trigger ALTER"}], + alter=True, ) rows = fresh_db.execute_returning_dicts("select * from test where i = 101") - assert [{"i": 101, "word": None}] == rows + assert [{"i": 101, "word": None, "extra": "Should trigger ALTER"}] == rows

def test_insert_ignore(fresh_db): ```

It can also return a dictionary with the following keys. This format is deprecated as-of Datasette 0.49 and will be removed by Datasette 1.0.

Documentation says that the old dictionary mechanism will be deprecated by 1.0:

https://github.com/simonw/datasette/blob/799ecae94824640bdff21f86997f69844048d5c3/docs/plugin_hooks.rst#L460

I'd be happy to accept a PR for this, provided it included updated unit tests that illustrate it working. I think this is a really good improvement.

This is deliberate behaviour, but I'm not at all attached to it - you're right in pointing out that it's actually pretty unexpected.

I'd be happy to change this behaviour so if you pass alter=True and then use .insert_all() on more than 100 rows it works as you would expect, instead of silently ignoring new columns past the first 100 rows. I don't expect that anyone would be depending on the current behaviour (ignore new columns after the first 100) such that this should be considered a backwards incompatible change.