On 2015-10-29 18:58, Walter Bright wrote:
The usual technique in the D test suite is to pipe the output to a file,
and then diff against what it should be. Yes, you wind up with about one
test file per error message.
https://github.com/D-Programming-Language/dmd/tree/master/test/fail_compilation
An alternative would be to have multiple tests in the same file, but the
actual source code for the tests in string literals. Then the testing
framework can write temporary files with the content from the string
literals and run the actual compilation on those temporary files.
I think that will work pretty good as long as the code inside the string
literals are fairly short.
Another advantage would be that any specific compiler flags that are
required for a given test can be passed as a separate argument instead
of being inline in the source code as a comment.
--
/Jacob Carlborg
