Nickolay Olshevsky

Could it be that some key server strips/changes this data as GnuPG doesn't seem to behave in this way?

Given that I didn't see any GnuPG-generated keys with such behaviour (and slight difference in default options, like absence of BZip2 compression) I guess this key was generated by some...

> @ni4 i agree with your underlying question. The specification suggests that what RNP is currently doing is the right thing for v6 signatures. I wouldn't want you to remove...

...Botan 3.3.0 has the following issue on macOS: https://github.com/randombit/botan/issues/3926

...Botan 2.19.4 doesn't have fix for the AEAD OCB issue: https://github.com/randombit/botan/issues/3812#issuecomment-1966273361

@antonsviridenko Sorry for a late followup. Another solutions could be putting the CFG_USERINPUTFD stuff in the RNP_RUN_TESTS define.

Thanks for spotting this! Should be related to multiple armored messages in one file. `rnp::ArmoredSource` with `AllowMultiple` flag should be used where dumper checks for armored input.

@antonsviridenko Thanks for noticing it, but it is not on the table until we implement and fully test v6 support.

@antonsviridenko Thanks for the suggestion! You are welcome to make PR implementing this :)

@ronaldtse what do you think? We already have separate repository for CMake modules, used in rnp: https://github.com/rnpgp/cmake-modules/ Actually never know/think before about upstreaming CMake modules to the CMake repository, but...