mhatipog opened a new issue, #12265:
URL: https://github.com/apache/arrow-rs/issues/12265

   fastparquet writes the empty `row_groups` list of FileMetaData with the 
one-byte header 0x00. #8826 reported this and #8827 accepted it; the list 
validation added in #9924 checks the element type even when the list is empty, 
so these files are rejected again.
   
   ```python
   import pandas as pd, fastparquet, datafusion
   fastparquet.write("/tmp/w1.parquet", pd.DataFrame({"k": pd.array([], 
dtype="Int32")}))
   ctx = datafusion.SessionContext()
   ctx.register_parquet("t", "/tmp/w1.parquet")
   ctx.sql("SELECT count(*) FROM t").show()
   # DataFusion 55.0.0: Parquet error: Expected list element type of Struct but 
got Byte. 54.0.0 reads it.
   ```
   
   Files with rows from the same writer open fine. Skipping the element-type 
check for size-0 lists would restore the old behaviour. (Polars fixed the same 
thing in pola-rs/polars#27749.)
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]

Reply via email to