JosiahWI
commented
3 months ago The motivation for the hacky workaround attempt that had less than desirable side effects was a bug in CMake's FindImageMagick module. Coincidentally, I made out the patch for that bug to CMake just this morning before becoming aware that webp_transform had been broken by the workaround.
https://gitlab.kitware.com/cmake/cmake/-/merge_requests/9617
Unfortunately we won't be able to depend on the CMake version containing the fix, once they release it, for some time, probably.
cmcfarlen
A bug was introduced in #10967 causing us not to register the auto_option for webp_transform if the ImageMagick::Magick++ target already exists. None of the other non-experimental plugins have auto_options at this time, so this demotes webp_transform to be always build if its dependency is found, in line with those other plugins.