Michael, James, et. al.,

When the SAS views are built using the script "data_step_view_prep.sas", the 
connection (or libname) to the RDBMS database in which the SAS views depend on, 
shows up in the SAS Server List tree panel (in the bottom left panel of SAS EG 
client)

Servers > SASApp > Libraries > {SAS Views database}

Our libname shows up in that tree as "NZDATA" as we are running on Netezza.  If 
you are running on Oracle, your libname under the Libraries branch in that tree 
will be named ORACDATA, if you didn't modify the SAS code.  SQL Server would 
probably have a different name.

This is all well and good, but I'm sure many of you have noticed some goofy 
things happen in the SAS EG client where you might get the following error on 
certain runs, and not on others... or you run the DCQ query and 6 hours later, 
you notice an issue that should have been transparent.


ERROR: Libname NZDATA is not assigned.
ERROR: Failure loading view PCORDATA.HARVEST.VIEW. Error detected during View 
Load request.



There's one line of code I'd like to suggest for the SAS script called 
"run_queries.sas" OR for any SAS scripts Duke plans to create or modify in the 
future (the DCQ query being one example), to avoid this issue for their 83+ 
sites who are submitting PCORI data.

If you do not run "data_step_view_prep.sas" and then run "run_queries.sas" 
while still in the same Microsoft Windows session for the SAS EG client task, 
that RDBMS libname will not show up, and therefore will cause that error to 
result.  As you may have noticed, SAS is quite buggy, in that you sometimes 
need to re-open the client for various reasons.  If that's the case, it 
wouldn't make sense to rebuild views if you can just redefine a connection that 
it needs.  So that's why I'd like to suggest Duke add this line of code, so you 
can open a fresh window of SAS EG, and not have to  worry about source code 
that was run separately from the main program.  By leaving this line of code 
out, we're just asking for problems.

I'm sure this line of code could be simplified with "libname {RDBMS} 
{variable}", but I'm just showing an example below.

libname nzdata netezza server='lab-netezza01.uwhealth.wisc.edu' 
database='QA_ETL_CDM' user='I2B2_INTELLAQ' password='secret' access=readonly;

********************************************************************************
* Submit data characterization query program
********************************************************************************;
%include "&qpath.infolder/data_characterization_query.sas";



_______________________________________________
Gpc-dev mailing list
[email protected]
http://listserv.kumc.edu/mailman/listinfo/gpc-dev

Reply via email to