Configurable naming of database columns
Opened this issue · 1 comments
Add an optional prefix pset_
(such as run(..., pset_prefix="pset_")
) to all pset variables such that we have a clear distinction in the db:
book keeping
_pset_id
_run_id
_pset_hash
...
pset content
pset_foo
pset_bar
...
results added by worker() and/or eval scripts. Users can here also add
prefixes as they wish, e.g. `postproc_` or `eval_`.
baz
boing
...
Another option is a minimal naming convention where "prefix" fields such as _pset_id
are added before calling a worker, which we do now. Then, users could be encouraged to name all returned fields in worker with a "postfix" notation such as result_
(inspired by how sklearn names estimator attrs after calling fit()
). Then, all fields which don't have either an underscore pre- or postfix are considered pset params and will be used in hash calculations, should we need to re-calculate the hash of a database row. Still, the refresh tooling may have an additional option to specify the included fields, as outlined in #15.