Closed timsaucer closed 1 month ago
@Michael-J-Ward It looks like we have a potential regression between 37.1.0 and 38.0.0. Namely substr
on 37.1.0 would accept a start and length, the parameters that should apply to substring
. I was using substr
incorrectly but it worked by accident on 37.1.0 because substr
and substring
were calling the same underlying function. In 38.0.0 this is updated and substr
now fails if you pass it two parameters.
This PR includes exposing substring
which was missing on the python side.
Which issue does this PR close?
Closes #696 Closes #712
Rationale for this change
This PR sets up a work flow to generate TPH-C 1Gb data set in CI, runs the 22 examples, and compares their results to the known answer file. By adding this PR we improve the robustness of our test suite.
What changes are included in this PR?
This PR adds the following changes:
Are there any user-facing changes?
substring
function exposed in python.Additional context
This PR replaces https://github.com/apache/datafusion-python/pull/710 which contains a lot of intermediate testing steps. This MR should be cleaner to review.