Hi everyone, is it possible to pass credentials of...
# questions
l
Hi everyone, is it possible to pass credentials of the underlying dataset when using it with CachedDataSet? e.g.
Copy code
test:
  type: CachedDataset
  versioned: true
  dataset:
    type: pandas.JSONDataSet
    filepath: ...
    load_args:
      lines: True
    credentials: ...
doesn’t work but this works
Copy code
test:
  type: pandas.JSONDataSet
  filepath: ...
  load_args:
    lines: True
  credentials: ...
I thought this was working at some point? I might be hallucinating though. Just want to double check quickly before I create my own CachedDataSet
1
👀 1
Sorry my bad. I included versioned: true and that tried to glob through the directory which my credentials didn’t have access to
👍🏼 1
n
Actually I don’t know if
versioned
work with any of these wrapper dataset like
PartitionedDataSet
l
The official doc includes versioned: true there for CachedDataSet
👀 1
n
Just to clarify it is only a permission issue, versioning still works right? or is this a doc issue.
l
• My issue was a permission issue. I turned off versioning and it works. I haven’t tested versioning yet. • My comment about versioning was in the doc was because you mentioned
don’t know if
versioned
work with any of these wrapper dataset like
PartitionedDataSet
If that’s the case, we should probably remove that from the doc for CachedDataSet. I haven’t verified that claim though
👍🏼 1
n
Cool, just to report back • CacheDataset
versioned
flag does work •
versioned
is not supported for
PartitionedDataset
🆒 1