unknown type OQS_SIG in sig_stfl.h included in oqs.h

[lingxinh-aws]

Describe the bug
"#include <oqs/sig_stfl.h>" was added in oqs.h in 971773a. It caused an error of "unknown type name 'OQS_SIG' " when compiling oqs-provider/oqsprov/oqs_sig.c, which includessig.h, which includes oqs.h. Then, OQS_SIG is referred in sig_stfl.h, but it is defined after #include <oqs/oqs.h> in the body of sig.h, so compilation failed.

To Reproduce
Steps to reproduce the behavior:

1. Git clone the oqs-provider
2. Run: cd oqs-provider
3. Run: scripts/fullbuild.sh

Expected behavior
OpenSSL, liboqs and oqs-provider are all successfully built.

Screenshots

Environment (please complete the following information):

• OS [ Amazon Linux 2 ]
• OpenSSL version [3.3.0-dev]
• Compiler version used [gcc 7.3.1]
• Build variables used ["-DCMAKE_BUILD_TYPE=Debug"]
• liboqs version [main branch]

Additional context
Add any other context about the problem here.

[baentsch]

Thanks for this report. Problem permeates through the complete OQS stack, see open-quantum-safe/oqs-provider#427 and open-quantum-safe/oqs-demos#281 . Possible work-around discussed here. A better solution of course is needed, see comment here: downstream-triggering CI should have caught this; still wonder why it didn't (ideas @SWilson4 ?)

[SWilson4]

Thanks for this report. Problem permeates through the complete OQS stack, see open-quantum-safe/oqs-provider#427 and open-quantum-safe/oqs-demos#281 . Possible work-around discussed here. A better solution of course is needed, see comment here: downstream-triggering CI should have caught this; still wonder why it didn't (ideas @SWilson4 ?)

I believe that it would have, but I think we (I really, as I was the one who merged the PR) simply missed manually triggering the downstream tests before merge. The trigger was broken during the final stages of the review, but I believe it was fixed before merge, so the tests should have been run (manually).

[SWilson4]

Resolved by #1820.