- fixed mpi and hpc builds, added file
* fix-mpi-build.patch
- Fixed python3.8 build and logging with the the two patches
* Fix-build-failures-for-python-3.8.patch which fixes the build
* Fix-TensorFlow-on-Python-3.8-logger-issue.patch which fixes the logging
- updated to latest stable v1 version 1.15.2 which fixes
CVE-2019-16778 (bsc#1159338)
- Highlights of this version are:
* MKL-DNN contraction kernels on by default
* Fixes a security vulnerability where converting a Python string
to a tf.float16 value produces a segmentation fault (CVE-2020-5215)
see RELEASE.md for full changelog
- updated following sources:
* tensorflow-1.15.2.tar.gz
* bazel-toolchains.tar.gz
* eigen.tar.gz
* gemmlowp.zip
* google-cloud-cpp.tar.gz
* license.rst.txt
* rules_closure.tar.gz
- added following patches:
* removed-docker-tools.patch to remove docker build during package build
* right-json-location.patch to make syslibs point to json
* tensorflow-make_aws_sdk_work_on_aarch64.patch
* fix-lite.patch
* json-feature-name.patch
- removed follwing sources
* abseil-pypi-v0.2.2.tar.gz
* aws-sdk-cpp-1.3.15.tar.gz
* boring_ssl.tar.gz
* double_conversion.zip
* fft.tgz
* flatbuffers_v1.9.0.tar.gz
* gast-0.2.0.tar.gz
* google-flatbuffers-1.10.0~pre.tar.gz
* google-nsync-1.20.1.tar.gz
* python-license.txt
* unicode-org-icu.tar.gz
- removed following patches:
* fix_mvapich_mpi_bzl.patch
* grpc-namespace-corrections.patch
* remove-keras.patch
* tensorflow-fix_lite.patch
* support-new-bazel.patch
- Suppress library-without-ldconfig-* in rpmlintrc as it seems
intentional and is currently causing builds to fail