Conversation
Codecov Report
@@ Coverage Diff @@
## master #32 +/- ##
============================================
+ Coverage 89.35% 89.39% +0.03%
+ Complexity 266 264 -2
============================================
Files 42 43 +1
Lines 1137 1150 +13
Branches 64 64
============================================
+ Hits 1016 1028 +12
Misses 86 86
- Partials 35 36 +1
Continue to review full report at Codecov.
|
| * Returns a set of applicable Extractors | ||
| */ | ||
| public ImmutableSet<ColumnLineageExtractor> buildExtractors() { | ||
| return outputColumns() |
There was a problem hiding this comment.
The extractor types and ordering is still necessary - this has just been temporarily disabled. One question is how extractors work in order (or nested?). An insert with SET col = CONCAT(colA, colB) will be parsed correctly with colA and colB properly set as parents of col, but the evaluation occurs after the DML statement evaluation and the type is a '$insert'. I think this just needs some extension to allow the ordering of extractors to change, as well as allowing an extractor registered on the table itself instead of columnType (I think this is necessary for DML, but still wrapping my head around it). Any advice appreciated.
2 participants
This is another WIP draft of partial DML support. The column->extractor selection is broken due to tables resolving to their names rather than the query type $query/$aggregate/etc.... This patch removes the extractor filtering, so all extractors will always run. I'm not sure if this is inefficient, b/c the visitors of irrelevant extractors don't seem to be invoked. Any thoughts appreciated.