Jae Anne Bach Hardie

If the data is reliably under 4000 bytes a `CAST(column AS nvarchar(4000))` on the select should work. If it's not I'm afraid you're out of luck until that issue in...

Nice work! We'll upgrade and hopefully the unixodbc dependency isn't too much trouble for people.

Unfortunately supporting additional column types is quite challenging since it would require changes to the driver being used, either in the form of updates to Erlang's ODBC driver or the...

I've just come back from a month's holiday and things are a bit less hectic now than they were for the first half of the year so I'm going to...

Does setting `returning: false` on the `insert/2` call allow the query to complete? I think that might be the best we can do, allow the user to turn off `OUTPUT`....

I meant `Ecto.Repo.insert/2` which is how the end-user interacts with the adapter. I also think I got it wrong and "return nothing" is expressed as `returning: []` rather than `false`,...

Thank you very much for tackling this. Submit a PR when you're ready and I'll prioritise reviewing and merging it.

Yeah, the Erland ODBC application has very limited support for non-standard column types. You would need to either patch Erlang OTP or write your own driver to use instead of...

The immediate use case is storing user preferences. Pretty rudimentary things like letting users choose whether to show or hide specific options. Any simple key-value store would be sufficient

I guess it depends on the exact amount of latency. If it's not very fast then having to load preferences from it on every single view, every time, might lead...