On Sat, 16 Jul 2022 14:12:56 +0200, Stephen Kitt <skitt@debian.org> wrote: > On Wed, 06 Jul 2022 09:52:41 +0800, Paul Wise <pabs@debian.org> wrote: > > On Wed, 2022-07-06 at 09:27 +0800, Bo YU wrote: > > > To my surprise this definition doesn't seem to take effect, > > > Either define or undef can build riscv64 package here. > > > > Probably the code that uses the definition is not adequately tested > > during the package build, if it was then there should be failure. > > That’s right. > > widl is a targetted tool anyway; it shouldn’t actually care about the host > CPU, the only thing that matters it the target CPU. I’ve been meaning to fix > this for a while, I guess it’s time to do it... Scratch that, it’s a little more complex; what matters in this particular case is whether the endianness of the host CPU matches the endianness of the target Windows ABI (which is always little-endian). Regards, Stephen
Attachment:
pgpxZi0tizbcQ.pgp
Description: OpenPGP digital signature