Run rdmd testsuite properly

[wilzbach]

Follow-up to #239 (comment)

This should make (1) it a lot easier to run the test suite properly (after all it's now part of the test Makefile target) and (2) should be a good step into debugging why Travis build succeed.

Currently, however, it fails:

pmake test_rdmd
../dmd/generated/linux/release/64/dmd -I../druntime/import -I../phobos -L-L../phobos/generated/linux/release/64 -m64 -w -de -ofgenerated/linux/64/rdmd_test rdmd_test.d
../phobos/std/parallelism.d(3523): Warning: statement is not reachable
../phobos/std/parallelism.d(3523): Warning: statement is not reachable
make: *** [posix.mak:72: generated/linux/64/rdmd_test] Error 1

I have looked into fixing rdmd_test two weeks ago and the fix for this is still in the Phobos queue: dlang/phobos#5502

CC @CyberShadow

[dlang-bot]

Thanks for your pull request, @wilzbach!

Bugzilla references

Your PR doesn't reference any Bugzilla issue.

If your PR contains non-trivial changes, please reference a Bugzilla issue or create a manual changelog.

[andralex]

@wilzbach thx!

[wilzbach]

@CyberShadow any idea why this happens on the for MODEL=32?

Argh - it seems like we need to rework the testsuite to include $MODEL

[wilzbach]

Wohoo this is finally passing on Travis 🎉
I had to explicitly pass the model in the rdmd_test suite though - @CyberShadow @andralex are you okay with this?

[CyberShadow]

Is there any instance of compilerSwitch that's not accompanied by an instance of of modelSwitch? If no, then just append model to compilerSwitch.

[wilzbach]

Is there any instance of compilerSwitch that's not accompanied by an instance of of modelSwitch?

Nope, but as execute excepts an array, we can't do ["dmd", "--compiler=foo -m32", ...].
Btw (a bit OT): is there any reason that execute doesn't allow ranges?

[CyberShadow]

You can just break the array and concatenate sub-arrays.

Anyway I looked and rdmdApp always occurs together with compilerSwitch, so I'd refactor those into an array, so that the invocation looks like execute(rdmdArgs ~ [...])

[wilzbach]

so I'd refactor those into an array, so that the invocation looks like execute(rdmdArgs ~ [...])

Fair enough. Done.

[CyberShadow]

I'd suggest pinning Digger to a specific version (as with DScanner) to avoid the breaking changes in Digger's master from affecting tools.

[wilzbach]

Okay - will do this in another PR