If it is de facto deprecated, then could it not just be deprecated with a comment stating why it has not been deleted? Calcite seems to have several API that are defacto deprecated and there is no way of knowing by reading the code. This would personally have made me much simpler for myself to get ramped on the calcite code base.
James On Wed, Jun 2, 2021 at 12:07 PM Julian Hyde <jh...@apache.org> wrote: > Master is a moving target. Apparently you are using a version of the > code from sometime in March. These links work better: > > [1] > https://github.com/apache/calcite/blob/796675c9b33e0461bc45a72780162d474a4b098b/core/src/main/java/org/apache/calcite/sql2rel/SqlToRelConverter.java#L2881 > > [2] > https://github.com/apache/calcite/blob/796675c9b33e0461bc45a72780162d474a4b098b/core/src/main/java/org/apache/calcite/sql2rel/SqlToRelConverter.java#L6209 > > I don't know the exact cause of what you are seeing, but when you set > "expand=false" you are choosing a newer (and better) code path. We do > not have feature parity between the code paths. Nor do we write tests > for both code paths. > > Some day I'd like to officially deprecate, then obsolete, > "expand=true". It is de facto deprecated because we put more effort > into the "expand=false" path. > > Julian > > > On Fri, May 28, 2021 at 1:19 AM stanilovsky evgeny > <estanilovs...@gridgain.com> wrote: > > > > Hi, calciters ) > > > > I am trying to figure out why DeduplicateCorrelateVariables [1] is called > > only if withExpand [2] flag is true ? Why we don`t need to deduplicate in > > appropriate case ? > > > > [1] > > > https://github.com/apache/calcite/blob/master/core/src/main/java/org/apache/calcite/sql2rel/SqlToRelConverter.java#L2881 > > > > [2] > > > https://github.com/apache/calcite/blob/master/core/src/main/java/org/apache/calcite/sql2rel/SqlToRelConverter.java#L6209 > > > > Thanks ! >
