Win10下基于CPU的Tensorflow编译和c++/python调用——问题汇总

安装请参考：Win10下基于CPU的Tensorflow编译和c++/python调用_caoyin1234的博客-CSDN博客

错误解决：
1. 遇见MSYS2卡66%==> Updating trust database.......
解决办法：关闭BIOS中SVM、或断网、或再次原目录覆盖安装；（可能原因清华源所致。）

2. ERROR: Analysis of target '//tensorflow/tools/pip_package:build_pip_package' failed; build aborted: error running 'git fetch origin refs/heads/*:refs/remotes/origin/* refs/tags/*:refs/tags/*' while working with @io_bazel_rules_docker:
打开WORKSPACE在以下添加：
# Download the rules_docker repository at release v0.14.3
load("@bazel_tools//tools/build_defs/repo:http.bzl", "http_archive")
http_archive(
    name = "io_bazel_rules_docker",
    sha256 = "6287241e033d247e9da5ff705dd6ef526bac39ae82f3d17de1b69f8cb313f9cd",
    strip_prefix = "rules_docker-0.14.3",
    urls = ["https://github.com/bazelbuild/rules_docker/releases/download/v0.14.3/rules_docker-v0.14.3.tar.gz"],
)

3. fatal error C1007: 无法识别的标志“-ReducedOptimizeHugeFunctions”(在“p2”中)
VS版本过低，最低要求Visual Studio 2019 版本 16.4 。（如果大版本相同，在vs中选择版本更新）

4.python -m pip install --upgrade pip时提示：Could not fetch URL https://pypi.tuna.tsinghua.edu.cn/simple/pip/: There was a problem confirming the ssl certificate: [SSL: CERTIFICATE_VERIFY_FAILED] certificate verify failed (_ssl.c:749) - skipping
Requirement already up-to-date: pip in c:\users\hyzk09\appdata\local\programs\python\python36\lib\site-packages
You are using pip version 9.0.1, however version 22.1 is available.
You should consider upgrading via the 'python -m pip install --upgrade pip' command.

解决：python -m pip install --upgrade pip  --trusted-host pypi.tuna.tsinghua.edu.cn

5.找不到"windows.h"

解决：pip install lxml --upgrade

6...\2019\Enterprise\VC\Tools\MSVC\14.29.30133\include\complex(672): error C2039: "copysign": 不是 "`global namespace'" 的成员”

解决：vs2019与编译版本不匹配造成，使用vs2017直接替换即可。也可以从以下链接自行下载：

VS2017对应目录下的complex文件-WindowsServer文档类资源-CSDN下载

7.严重性    代码    说明    项目    文件    行    禁止显示状态
错误    LNK2001    无法解析的外部符号 "public: __cdecl tensorflow::TensorShapeBase<class tensorflow::TensorShape>::TensorShapeBase<class tensorflow::TensorShape>(class absl::lts_20210324::Span<__int64 const >)" (??0?$TensorShapeBase@VTensorShape@tensorflow@@@tensorflow@@QEAA@V?$Span@$$CB_J@lts_20210324@absl@@@Z)    tf_test    D:\WorkSpace\WorkGroup\DeepLearning\DEMO\tf_test\tf_test\tf_test\tf_test.obj    1

解决办法：进入tensorflow-master\tensorflow\tools\def_file_filter，编辑def_file_filter.py.tpl文件，将错误按下格式添加：

# Header for the def file. (找到这一行代码)
    if args.target:
    def_fp.write("LIBRARY " + args.target + "\n")
    def_fp.write("EXPORTS\n")
    def_fp.write("\t ??1OpDef@tensorflow@@UEAA@XZ\n")
    # 下面两个就是复制的错误信息
    def_fp.write("\t ?NewSession@tensorflow@@YA?AVStatus@1@AEBUSessionOptions@1@PEAPEAVSession@1@@Z\n")
    def_fp.write("\t ??0SessionOptions@tensorflow@@QEAA@XZ\n")

注意：一定要格式对齐！修改后一定要重新生成！（以后出现同样的问题，解决办法雷同）