tustvold opened a new issue, #5283:
URL: https://github.com/apache/arrow-datafusion/issues/5283

   **Is your feature request related to a problem or challenge? Please describe 
what you are trying to do.**
   
   It is fairly common to get errors from DataFusion such as in 
https://github.com/apache/arrow-ballista/issues/665
   
   ```
   2023-02-12T22:26:10.297254Z ERROR tokio-runtime-worker ThreadId(02) 
ballista_scheduler::scheduler_server::query_stage_scheduler: Error planning job 
t9P5iXR: DataFusionError(SchemaError(FieldNotFound { field: Column { relation: 
Some("part"), name: "p_brand" }, valid_fields: [Column { relation: None, name: 
"COUNT(DISTINCT partsupp.ps_suppkey)" }] }))  
   ```
   
   Which do not make it easy to understand the context in which the error was 
encountered. Typically the approach I follow is to use an interactive debugger, 
such as GDB, to catch the error. However, this requires having a local 
reproducer, which isn't always possible.
   
   **Describe the solution you'd like**
   
   Backtraces are a commonly used mechanism to automatically provide context on 
where an error was encountered. Whilst they have issues, especially in async 
contexts, they are better than the current situation where diagnosing such 
issues is non-trivial.
   
   Adding a backtrace to the error variants that have corresponding `From` 
conversions, i.e. `ArrowError`, `ParquetError`, `AvroError`, `ObjectStore`, 
`IoError`, `SchemaError`, etc... would automatically capture the context in 
which this conversion is performed.
   
   APIs can then use `DataFusionError::find_root` to get the leaf variant, and 
from that extract the backtrace if any.
   
   This is predicated on the fact that `DataFusionError` is rarely used to 
implement control flow, and so the cost of obtaining a backtrace is only paid 
on exceptional paths.
   
   **Describe alternatives you've considered**
   
   **Additional context**
   Add any other context or screenshots about the feature request here.
   


-- 
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