feat(core): update to substrait v0.100.0 - #1082
Conversation
4fbc30c to
ecaaf51
Compare
|
I approved the workflow runs yesterday and it looks like the Spark dialect file also needs to be updated in this PR. The following command should regenerate it: |
|
@nielspardon done |
|
We fixed a bug in the Spark dialect generation code in #1086. If you rebase we don't need the Spark dialect changes anymore. |
ccc4658 to
fa9284c
Compare
|
@nielspardon done |
|
I think you still would need to drop that second commit with the spark changes |
fa9284c to
e20656e
Compare
|
@nielspardon try now |
nielspardon
left a comment
There was a problem hiding this comment.
Now that the generator drops variants whose signature has no Spark sample value, no unsigned function is emitted (Spark has no unsigned type mapping), so this alias has nothing referencing it: the rebased dialect has zero functions with source: "unsigned_integers". On main, declared and used aliases match exactly at 12/12, and functions_list/functions_set are precedent for a catalog extension that is simply not declared here. Suggestions below to drop both lines; they can come back together with the functions if Spark ever gains an unsigned type mapping.
nielspardon
left a comment
There was a problem hiding this comment.
LGTM, I think we can safely drop the two lines in the spark module
….scala remove `unsigned_integers` Co-authored-by: Niels Pardon <mail@niels-pardon.de>
remove unsigned_integers Co-authored-by: Niels Pardon <mail@niels-pardon.de>
Uh oh!
There was an error while loading. Please reload this page.