Results for TDVT test case STR([num2]) are correct but do not match any expected results

[Michael Habiger]

About You:
Name: Michael Habiger
Company: Actian Corporation

Your question:
When running the TDVT test suite against Actian X (v11.2) or Actian Vector (v6.3), the actual results for test case STR([num2]) do not match any of the expected results.

The reason for the lack of matching expected results is that actual results contain trailing zeros due to the Actian database default display format for float4 (real) and float8/float (double precision) being n10.3, i.e. the scale is 3, thus 10.98 is shown as 10.980, 11.5 as 11.500, and so forth.

Here are the Actian results for test case STR([num2]) showing the values with trailing zeros due to the default scale of 3 for float values.

%null%
"10.980"
"11.500"
"11.690"
"13.040"
"16.730"
"17.250"
"17.860"
"3.790"
"6.460"
"6.800"
"7.870"
"8.510"
"8.980"

The TDVT suite has 14 expected results files in this folder for test case STR([num2]) to match results from various vendors (db2,firebird,greenplum,hana,hyper,kognitio,memsql,oracle,postgres,postgres_jdbc,sqlserver,sybaseiq,teradata
) but none of the results match the Actian results due to the trailing zeros.

Request:
We would like to request that a new expectation file be added (or an existing one be updated?) to match the Actian results.

Actian TDVT Actual Results Files:
actual.setup.cast.str.txt
actual.setup.cast.str_diff.txt

[Luke Petschauer]

Internal tracking: W-14129458

[Ross Brown]

Thanks for raising this issue. It is a known pain point and we've elevated finding a more general solution in our backlog. We'll update this issue with future progress in this area.

In the meantime, you can skip this test in your runs with a note about this known issue.

[Michael Habiger]

Thanks for the explanation.