Uh oh!
There was an error while loading. Please reload this page.
BigQuery: Augment BQ Model metadata. - #5214
Conversation
This change introduces extended ML model metadata into the existing CRUD implementation for BigQuery ML models. This implementation exposes the underlying discovery types directly without an abstraction for the ML metadata. Methods that provide this metadata are all marked @BetaApi to indicate the getters may change.
Codecov Report
@@ Coverage Diff @@## master #5214 +/- ##
============================================
+ Coverage 50.41% 50.42% +<.01% - Complexity 23796 23799 +3
============================================
Files 2251 2251 Lines 226836 226874 +38 Branches 24966 24969 +3 ============================================
+ Hits 114370 114392 +22 - Misses 103858 103871 +13 - Partials 8608 8611 +3
Continue to review full report at Codecov.
|
tswast
left a comment
There was a problem hiding this comment.
LGTM. This is so much nicer than if we had to duplicate all those properties manually.
shollyman
commented
May 23, 2019
The combination of the annotation, that the representations are effectively read only, and the fact that we pin the discovery artifacts in the maven dependencies makes me much more comfortable here. Assuming no other comments come up, I'll merge it when I'm back in the office tomorrow my time. |
This is a breaking change to a @BetaApi within BigQuery. It swaps in a new veneer StandardSQLField type in place of the underlying discovery type for getting information about the feature and label columns from ML models, which was functionality introduced in: googleapis#5214
…ns. (#5732) This is a breaking change to a @BetaApi within BigQuery. It swaps in a new veneer StandardSQLField type in place of the underlying discovery type for getting information about the feature and label columns from ML models, which was functionality introduced in: googleapis/google-cloud-java#5214
This change introduces extended ML model metadata into the existing CRUD
implementation for BigQuery ML models. This implementation exposes the
underlying discovery types directly without an abstraction for the ML metadata.
Methods that provide this metadata are all marked @BetaApi to indicate the getters may change.