We might consider load() / unload() methods, but fully representing the semantics of https://docs.aws.amazon.com/redshift/latest/dg/r_COPY.html in Python seems ambitious (and a moving target...)
my only real issue here is about the handling of multiple queries where a query doesn't return any results -- it seems to me that the length of the result should be the same as the length of the list of queries -- otherwise how will we know which result corresponds to which query?
if users aren't intended to implement this, can it be _BaseRedshiftResource with a docstring explaining that
why a tuple?
might be nice to link via intersphinx here
should be List[Tuple] or List[Tuple[Any, ...]]
Is this metadata not sufficiently represented in the structured log entry? I would rather not introduce this kind of semistructured metadata in log lines
this is List[List[Tuple]] or List[List[Tuple[Any, ...]]]
shouldn't we still do results.append()?
hm, seems weird not to be able to config this from env
is this a little misleading? if i set autocommit: false afaict we don't disable anything -- it's just like this being None
maybe link to docs for acceptable values
should be a sphinx code-block -- literal include would be better yet as we could put it under test
you mean vs. a list? don't feel strongly, just have used this convention elsewhere for extracting args from config
yah agree, removed
yes we absolutely should, added
Yup! fortunately Alex just landed https://dagster.phacility.com/D2267 - updated to use IntSource
yes thanks for catching - I've updated the code to actually do what this text describes
thanks for flagging - yeah made this a code block for now, I'll do the literal include as a follow up