HyunWoo Oh

Also, in the Object Storage Engine, it is weird that there is a directory type and that file is not directory.

I found where that file location information come from! https://github.com/fsspec/filesystem_spec/blob/master/fsspec/spec.py#L97  

Thanks for replying, @Fokko! Yes :) I'm definitely interested in fixing the above. Yesterday, I found something related to the above log: However, is it our responsibility to handle that...

I added some PyIceberg code that retrieves file information from the Arrow file system to make it easier for you to find. [iceberg-python.pyiceberg.io.pyarrow.py](https://github.com/apache/iceberg-python/blob/main/pyiceberg/io/pyarrow.py#L262) ``` python def _file_info(self) -> FileInfo: """Retrieve...

Hi, @kevinjqliu :) Thank you for summarizing. > Looking at the screenshot. It looks like you're running create_table with a GCS table location. On table creation, the metadata json file...

I think this might be the clue that’s worth digging into further: https://github.com/apache/arrow/blob/main/cpp/src/arrow/filesystem/gcsfs.cc ``` c++ Result GetFileInfo(const GcsPath& path) { if (path.object.empty()) { auto meta = client_.GetBucketMetadata(path.bucket); return GetFileInfoBucket(path, std::move(meta).status());...

Sure @kevinjqliu :) Below the image, top of the logs is the return type. FileInfo for gs://seoul_data_v1/warehouse/test/metadata/00000-6f9631fe-18db-4743-915b-36d7562d4073.metadata.json : type=FileType.Directory

My ec2 is provisioned on GCP. Basically they set the gcloud account owner. So I can view the bucket inside.  my code is this. ``` python catalog.create_table("analytics.test", schema=arrow_table.schema, location="gs://seoul_data_v1/warehouse/test")...

@kevinjqliu location = 'seoul_data_v1/warehouse/test/metadata/00000-6f9631fe-18db-4743-915b-36d7562d4073.metadata.json'

@kevinjqliu There is nothing. I ran the gsutil ls command beforehand, and the seoul_data_v1/warehouse/test/ directory did not exist. As you know, 00000-6f9631fe-18db-4743-915b-36d7562d4073.metadata.json is a unique ID generated by PyIceberg. So,...