I wonder how hard we should try to build on 32-bit platforms.
Intuitively I think it should be a secondary or ancillary concern. Especially as we have a focus on large datasets and a 32-bit address space doesn't really help there. Regards Antoine. Le 13/01/2019 à 17:41, Wes McKinney a écrit : > Thanks Jeroen (you can reply on the vote thread about this next time > there's an issue with an RC). > > Unfortunately I don't think it would be a good idea to block the release > over this. We don't have CI for this build configuration, so we should try > to fill this gap for the 0.13 release > > On Sun, Jan 13, 2019, 7:55 AM Jeroen Ooms <jeroeno...@gmail.com wrote: > >> Did a test build of the 0.12 RC with mingw-w64 on GCC-8.2.0, getting: >> Build error: '_mm_crc32_u64' was not declared in this scope. Full log: >> >> https://ci.appveyor.com/project/jeroen/rtools-packages/build/job/c619b6ibu1uqeqln >> >> I'm not sure what this does but I would guess that you should be using >> _mm_crc32_u32 instead of _mm_crc32_u64 on 32bit (or disable SSE4 >> extensions all together)? >> >
