Stream: brlcad

Topic: add_test wrapping

starseeker (Nov 10 2020 at 16:17):

So the first thing to do is probably to replace add_test with brlcad_add_test in our code and see what breaks (if anything)

Sean (Nov 10 2020 at 16:20):

yeah, I suppose. should it be lowercase like that or uppercase for consistency? do we have anything else lowercase?

Sean (Nov 10 2020 at 16:21):

I see we have BRLCAD_REGRESSION_TEST() and friends in Targets

starseeker (Nov 10 2020 at 16:21):

Er, yeah - upper case.

starseeker (Nov 10 2020 at 16:21):

I guess we can keep that workaround, but arrgh...

starseeker (Nov 10 2020 at 16:22):

(different calling for all the space cases)

Sean (Nov 10 2020 at 16:22):

then in the function, it'd change from calling _add_test() to add_test()

starseeker (Nov 10 2020 at 16:22):

Was that space issue specific to add_test? If so, maybe we'd be good...

Sean (Nov 10 2020 at 16:30):

yes, it was iirc, specific to both add_test and _add_test, but not to our function calling them -- that's why the workaround worked. I had to backtrack to figure out whose code was ignoring empty args.

starseeker (Nov 10 2020 at 16:32):

OK, so BRLCAD_ADD_TEST may do the trick, and with ExternalProject_Add coming online any 3rd party tests we may care about will have to be handled differently anyway.

starseeker (Nov 10 2020 at 16:33):

Should I do the alteration in the extbuild branch, since that's where we've got ExternalProject_Add set up, or go ahead and do it in trunk?

Sean (Nov 10 2020 at 16:46):

I'd say trunk as it's not really specific to the external build

Sean (Nov 10 2020 at 16:47):

and then merge back to extbuild

want me to refactor trunk?

starseeker (Nov 10 2020 at 17:11):

If you've got time, but more urgent is probably to try extbuild on the Mac

starseeker (Nov 10 2020 at 17:13):

I'm hitting some odd-pathnames test issues (a couple of which I'm not really sure are resolvable - if the external build system doesn't like them...) but if the Windows issue really has been addressed then I'm closing in on merge.

starseeker (Nov 10 2020 at 20:22):

@Sean I switched the wrapper - getting a couple new bu_vls_simplify failures, but not sure why yet...

starseeker (Nov 10 2020 at 20:24):

Ah! that's the empty failure.

starseeker (Nov 10 2020 at 20:24):

/me tries to boil it down

Sean (Nov 10 2020 at 20:26):

I just commented again on the github issue. The chunk of code that packs and unpacks is probably still necessary unless there's some way to forward a subset of argv quoted.

Sean (Nov 10 2020 at 20:27):

what little digging I did indicates it's an issue with forwarding ${ARGV}'s in general

Sean (Nov 10 2020 at 20:27):

so we happened to hit it in our override function, but it's an issue for any function

starseeker (Nov 10 2020 at 20:44):

There we go - restored and working.

Last updated: Oct 09 2024 at 00:44 UTC