ngam
ngam
> Blocked by pyuv. does adding libuv as req help?
Still an issue: ``` clang -D_FORTIFY_SOURCE=2 -isystem $PREFIX/include -mmacosx-version-min=11.0 -ftree-vectorize -fPIC -fPIE -fstack-protector-strong -O2 -pipe -isystem $PREFIX/include -fdebug-prefix-map=$SRC_DIR=/usr/local/src/conda/openlibm-0.8.0 -fdebug-prefix-map=$PREFIX=/usr/local/src/conda-prefix -fno-builtin -fno-strict-aliasing -fPIC -std=c99 -Wall -I$SRC_DIR -I$SRC_DIR/include -I$SRC_DIR/aarch64 -I$SRC_DIR/src -DASSEMBLER -D__BSD_VISIBLE...
Oddly, this does NOT happen for our linux aarch64 [cross-compile](https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=436152&view=logs&j=6494c7a6-a019-59a3-6f3d-9bab8ca96717). Only MacOS [cross-compile](https://dev.azure.com/conda-forge/feedstock-builds/_build/results?buildId=436152&view=logs&j=ced5d8de-8227-5f3f-33a1-45cf1592c45a). Maybe something to do with SDK, etc.?
And, one more piece of info: Compiles totally fine natively (on mac arm64). ``` +build_alias=x86_64-apple-darwin13.4.0 +host_alias=arm64-apple-darwin20.0.0 ```` When building natively, `build_alias` will be `arm64-apple-darwin20.0.0`
As @isuruf fixed it in https://github.com/conda-forge/openlibm-feedstock/pull/29/commits/f83e0bae83c4245e11c76e54e77905e5ed85d101, this seems to resolve this particular problem for osx: `sed -i.bak "s/CC = clang/CC ?= clang/g" Make.inc`
@JackCaoG any more insight on this? opened a new issue https://github.com/tensorflow/tensorflow/issues/56430
Interesting. I wonder if reverting the contexpr changes can fix this... Waiting for the tensorflow people to weigh in... Thanks a lot, I made sure to link this issue in...
What compiler version are you both using? According to the response from upstream, https://github.com/tensorflow/tensorflow/issues/56430#issuecomment-1161289033, it may be a version issue. I am trying that now in conda-forge and it seems...
FYI, resolved by https://github.com/tensorflow/tensorflow/commit/bc4521dd193290f86bd5de8a56cefbcbfeae3213
@ocefpaf @jswhit any way we could help here? FWIW, this issue is *not* present on M1 Macs.