The project is done now.
In case anybody finds this thread – I have tried some things, but the SQL Server database (running on Windows) itself was problematic to work with.
mssqlex with and without mssql_ecto
tds with and without tds_ecto
- Ecto in version 2
:unicode.characters_to_list calls (strings are character lists in erlang)
https://github.com/tallakt/codepagex in an attempt to convert the data to utf-8 (I enabled the CP1252 encoding in
VENDORS/MICSFT/WINDOWS for this)
I was able to query the database with all adapters and drivers. The encoding issue surprisingly took most of the time. Most of the time I got recognizable text intermixed with garbage. The data could be queried correctly with raw SQL in SSMS (the SQL server GUI). There is another account of this difficulty with no solution here.
I suspect that at least one layer in the chain lied about its encoding.
I was not granted much time to try this. The database was not reachable off-premise, either.
Be that as it may, we ended up not using elixir.
Again, Thanks for everybody’s help.