yangshangqing95 commented on PR #17282:
URL: https://github.com/apache/iceberg/pull/17282#issuecomment-5122516401

   > `puffin` is the file format, i think it might be better to make the name 
more generic. maybe something like `statistics_files` and 
`all_statistics_files` https://iceberg.apache.org/spec/#table-statistics
   
   Hi @kevinjqliu Thanks for the comment! I see the distinction you are making: 
ideally, a metadata-table name should describe a logical Iceberg concept rather 
than an implementation detail of the current physical format.
   
   A truly format-neutral alternative could be something like `auxiliary_files` 
and `all_auxiliary_files`, together with an explicit `file_format` column. That 
would leave room for other container formats in the future.
   
   My concern is that “auxiliary file” is not currently a defined Iceberg 
concept and its scope would be ambiguous. It could potentially include 
statistics files, partition statistics files, encryption-related files, index 
files, or other metadata artifacts. Introducing that abstraction would 
therefore require a broader definition of what qualifies as an auxiliary file, 
rather than being only a naming change.
   
   In contrast, Puffin already has a clear and established role in Iceberg as 
the container format for blobs referenced by table metadata and manifests. The 
rows exposed by these tables represent Puffin containers directly, and the 
schema includes container-specific information such as referenced blob count, 
blob types, and referenced field IDs.
   
   Puffin is also not limited to table statistics. It is already used for 
deletion vectors and is designed to support additional blob types in the 
future. For this reason, `puffin_files` and `all_puffin_files` describe the 
current row entity precisely.
   
   Given that Puffin is already a well-defined Iceberg concept, I am not sure 
we gain much by introducing a new, broader abstraction solely to avoid naming 
the file format. Would you still prefer defining a format-neutral 
auxiliary-file abstraction, including its intended scope, or does keeping the 
tables explicitly focused on Puffin files seem reasonable?
   
   and for `puffin_files` https://github.com/apache/iceberg/pull/17192


-- 
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]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to