Hi Daner,
SD1082 rule algorithm is adjusted (2304.1 engine) to ignore any custom variables which are not defined in ADaM standard. In previous version any custom ADaM variables were considered as Char type, so SD1082 rule generates false-positive messages for numeric (new custom) variables.
Sorry for inconvenience.
Kind Regards,
Sergiy
Hello,
I'm getting this warning on (numeric) ANRLO and ANRHI.
Is that expected? I wouldn't think any numeric variable would be triggering this warning...
Thanks!
At least for the case that SAS-XPT is used, the algorithm should first check whether the variable is "char" or "num" in the header in the XPT file.
In case of "num", no check should be executed as in SAS-XPT, the length for numeric variables is fixed to 8 bytes (i.e. you can't change it).
For modern formats like Dataset-JSON (https://www.cdisc.org/dataset-json#standard__Specification) the information whether the variable is numeric or not can be either found in the header (Dataset-JSON) and/or in the define.xml (Dataset-XML). But essentially, with modern formats, there is no limitation to the variable length (200 in SAS-XPT) and no "wasting of bytes" (see e.g. http://cdiscguru.blogspot.com/2016/12/how-sas-xpt-works-well-inefficient.html), so there should not be such rule anyway.
Basing the check on the name of the variable is definitely the wrong approach.
Hi Alice,
SD1082 should not produce an error message for ANRLO, ANRHI variables.
Could you send me your sample dataset for issue diagnostics?
Thank you,
Sergiy
sergiy (at) pinnacle21 (dot) com
I'm getting this error message for numeric variables now. Do we need to fix the numeric variable length in ADaM now? Thanks