robertwb commented on PR #31558:
URL: https://github.com/apache/beam/pull/31558#issuecomment-2166214597

   Taking a step back, is the problem that you're trying to solve is that those 
transforms that come from a `[Typed]SchemaTransformProvider` we already have a 
`(identifier + configRow) -> [Schema]Transform` mapping, and we should be able 
to automatically derive the inverse `[Schema]Transform -> (identifier + 
configRow)` mapping for these transforms?
   
   In this case, it's a bit odd because it seems we can only derive the inverse 
for the specific instances that we created, not generally for all instances of 
that class. (But maybe that's good enough sometimes?) Might be cleaner in that 
case to just have a separate (weakref) mapping (registrar) from PTransform 
instances to their corresponding configs rather add a register method and 
private members to `[Schema]Transform` itself. 
   
   Separately it'd be good to make defining this pair of mappings so easy that 
it's "just how users write PTransforms" (maybe working it into builder patterns 
or providing a good annotation or something like that) and we get their 
externalization and semantic graph representation for free. But this is likely 
a larger project. 


-- 
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