dwsmith1983 opened a new pull request, #7561:
URL: https://github.com/apache/datafusion-comet/pull/7561

   ## Which issue does this PR close?
   
   Part of #7128.
   
   This is the second of four pull requests for #7128, each covering one area. 
They build on #7127, which adds the URL redaction helpers they use, so #7127 
merges first. Suggested order after it:
   
   1. #7556: object store setup and its error messages.
   2. This one: the Parquet writer's URL errors, plan display and debug log.
   3. Native Delta's URL display and deletion vector errors.
   4. The rejected path reason on the JVM side, which closes #7128.
   
   They target `main`, so until #7127 merges each one also shows its commits.
   
   ## Rationale for this change
   
   The native Parquet writer prints its output path in its URL errors, its plan 
line and the debug line it logs for each file. A path can carry a password in 
its userinfo or a SAS token or signature in its query string, so those lines 
put secrets in task failures and logs.
   
   ## What changes are included in this PR?
   
   - The writer's errors, plan line and per-file debug line show the path 
without its password, query or fragment, and outside ABFS without its username. 
ABFS keeps the username, since it names the container.
   - A path with nothing to hide is shown exactly as Spark passed it, so local 
writes read as before.
   - A path that does not parse shows only its scheme.
   
   The HDFS store setup and operator messages also print the error they wrap; 
for a URL with no host that error quotes the URL until #7556 lands.
   
   ## How are these changes tested?
   
   Native tests run the writer's error paths, its plan display and its debug 
log with a path holding a password, a username, a signature query and a 
fragment, and check that none of them appears while the scheme, host and path 
do. A table test covers each part that gets removed, the ABFS container, 
unparseable paths, and local paths that must stay as they are. Each test fails 
on the base branch. The two HDFS messages need a running name node, so no unit 
test reaches them.
   


-- 
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