The move prevents some errors though, all the errors cannot be gone. For example, in o.a.s.sql.catalyst.analysis.*Suite,
The case ' https://github.com/maropu/spark/commit/961b5e99e2136167f175598ed36585987cc1e236 ' causes 3 errors. AnalysisSuite: - analyze project *** FAILED *** AnalysisErrorSuite: - unresolved attributes *** FAILED *** - unresolved window function *** FAILED *** The case '' https://github.com/maropu/spark/commit/da0e1c4da9960150770c1c819cc6c092a9849cb7 " causes a 1 error. AnalysisErrorSuite: - unresolved window function *** FAILED *** I'm currently looking into this tough, not sure about culprits. Thanks, // maropu On Sun, Jul 19, 2015 at 2:40 AM, Ted Yu <yuzhih...@gmail.com> wrote: > What if you move your addition to before line 64 (in master branch there > is case for if e.checkInputDataTypes().isFailure): > > case c: Cast if !c.resolved => > > Cheers > > On Wed, Jul 15, 2015 at 12:47 AM, Takeshi Yamamuro <linguin....@gmail.com> > wrote: > >> Hi, devs >> >> I found that the case of 'Expression.resolved != >> (Expression.childrenResolved && checkInputDataTypes().isSuccess)' >> occurs in the output of Analyzer. >> That is, some tests in o.a.s.sql.* fail if the codes below are added in >> CheckAnalysis: >> >> >> https://github.com/maropu/spark/commit/a488eee8351f5ec49854eef0266e4445269d5867 >> >> Is this a correct behaviour in catalyst? >> If correct, anyone explains the case if this happens? >> >> Thanks, >> takeshi >> >> -- >> --- >> Takeshi Yamamuro (maropu) >> > > -- --- Takeshi Yamamuro