jvanstraten commented on PR #12916:
URL: https://github.com/apache/arrow/pull/12916#issuecomment-1105800137

   > A single file shouldn't have different URIs referring to it.
   
   I guess I was mainly thinking from the validator's perspective, where a URI 
is just something it resolves to a YAML file, in which case it's not unique at 
all. Then I started thinking about relative references, then about awkward 
versioning stuff (if we always have to use the exact same URI for Arrow's 
types, how can we ever change something without immediately wrecking everything 
previously out there? And if they're versioned, what's the granularity?), and 
then gave up thinking about it. I do agree that we'll need inter-YAML 
references at one point or another, though, and mandating that two URIs must be 
identical for them to be treated as the same extension is probably the right 
approach, unless the manner of referencing things is changed considerably.
   
   Anyway, it seems like you've thought about this in terms of compatibility 
across implementations a lot more than I have, which is ultimately more 
important, and what you're saying sounds reasonable to me.
   
   > I think this file should also contain kernels for all of the standard 
Substrait functions (e.g. `add_uint8_uint8`).
   
   Nitpicking, but I'm not sure what you mean by kernels here (did you mean 
prototypes?), and the function name would be just `add` in this case, with the 
overload referred to in the plan as `add:u!uint8_u!uint8` (I'm really only 
pointing it out because this part of the spec is not very obvious. The relevant 
bit is 
[here](https://substrait.io/extensions/#function-signature-compound-names); I 
missed it completely until Jacques pointed it out to me on slack.)


-- 
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