The CI program scans this repository for all files ending in .dba
.
The filename encodes how the result should be interpreted. c
means
"compile", r
means "run", y
means "yes" and n
means "no".
- To assert a snippet must compile on both odbc and dbpc, the file should
start with
cy-
- To assert a snippet must fail to compile on both odbc and dbpc, the file
should start with
cn-
- To assert a snippet must compile on odbc but fail to compile on dbpc, the
file should start with
odbc-
- To assert a snippet must compile on dbpc but fail to compile on odbc, the
file should start with
dbpc-
If a file is asserted to compile successfully, then we can additionally specify what the runtime behavior should be.
- To assert a snippet's output must be identical on both odbc and dbpc, the
file should start with
cy-ry-
- To assert a snippet's output must be different between odbc and dbpc, the
file should start with
cy-rn-
If the snippet only compiles on one compiler, or if the runtime behavior is
different between odbc and dbpc, then we must also specify the expected output
by creating an identically named file ending in either .out
or in the
case of multiple different behaviors, .dbpout
and .odbout
. For
example, if our snippet should only compile with odbc, then we would create
the two files:
- odbc-description of test.dba
- odbc-description of test.out
If our snippet compiles on both but has different runtime behavior, then we would create the three files:
- cy-rn-description of test.dba
- cy-rn-description of test.dbpout
- cy-rn-description of test.odbout
Everything after the initial encoding (and before the file extension) should be a description of what the snippet is trying to test. This is usually just a sentence with spaces between each word.