hash_match: terminates in unexpected error in some cases when called on DST

[thomas-rasmussen]

An example has been found where the hash_macro terminates in unexpected errors during the iterations of matching when used on the DST servers. The error seems to be related to the proc append steps in line 816-822.

• ~Does the macro work at all on DST? If yes, then error is specific to the input data.~ It works on DST.

• ~Anders has used the macro on either DST or SDS. Find out where.~ DST

• If error is data-specific, replicate problem and make sure that test is added to test syntax.

• If the error is data-specific the problem is probably because the first strata is weird in some way that the macro does not account for.

• ~If error is not data-specific it is probably caused by a SAS system option setting that differs on DST from our local server where the macro was developed making proc append behave differently.~ Must be a data-specific problem it seems like.

[thomas-rasmussen]

Issue should be fixed in the issue-7 branch. Waiting for feedback before merging to master branch.

[thomas-rasmussen]

The solution fixed a bug, but not actually the one that as the problem. The problem seems to be that (in some cases) proc append does not work on DST, if the base dataset does not exist, as it won't be in the first iteration of the macro. Maybe it IS some system option that is (sometimes?) different on the DST SAS server. Another way to fix the problem might be to not use proc append in the first iteration, and instead explictly define the dataset used as a base in the rest of hte iterations.

[thomas-rasmussen]

issue has been fixed