Consideration of Type Safety

[JakkuSakura]

Are there any considerations/plans regarding type safety?

Sometimes, a slight mismatch between the DDL and rust struct will cause strange errors. If we add checks about type safety, these errors could be solved more easily

[stepancheg]

Yes. Errors look like this:

Code: 33. DB::Exception: Cannot read all data. Bytes read: 3. Bytes expected: 8.: (at row 1)\n: While executing BinaryRowInputFormat. (CANNOT_READ_ALL_DATA)

Unhelpful, hard to debug.

And prevents schema migration. For example, if you database type changed from non-null string to nullable string, it is not possible to write an inserter code which handles both old and new versions of the schema.

And can lead to insertion of garbage into table if accidentally the number of bytes will match.

[slvrtrn]

We are looking into possibilities of using RowBinaryWithNamesAndTypes as the default format, so there will be stronger safety guarantees and better errors in case of struct mismatch with DDL. Can't give any ETA yet, as this is currently on the draft stage, and we need to be sure that we get the implementation right.

[slvrtrn]

Here's a draft PR that adds RowBinaryWithNamesAndTypes support on Query for stricter safety guarantees: #221. Your feedback is welcome!