引言： Tensorflow大名鼎鼎，这里不再赘述其为何物。这里讲描述在安装python包的时候碰到的“No matching distribution found for tensorflow”，其原因以及如何解决。

简单的安装tensorflow

这里安装的tensorflow的cpu版本，gpu版本可以自行搜索安装指南，或者参考如下指令：

pip3 install tensorflow #cpu

这里使用的python 3.6.3版本。

pip3 install tensorflow-gpu

这里是gpu的版本。

window的环境

window 7.

问题描述：

pip3 install tensorflow

如此简单的指令，应该不会出错吧，结果得到如下错误信息：

Collecting tensorflow
Could not find a version that satisfies the requirement tensorflow (from versions: ) No matching distribution found for tensorflow

为什么没有找到tensorflow呢？那我们自行找找看吧？

pip3 search tensorflow

具体的输出信息如下：

…………………..
tensorbase (0.3) - Minimalistic TensorFlow
Framework
tensorbayes (0.3.0) - Deep Variational Inference in
TensorFlow
tensorflow-tensorboard (0.4.0rc3) - TensorBoard lets you watch
Tensors Flow
tensorboard_logger (0.0.4) - Log TensorBoard events without
Tensorflow
tensorboardX (0.8) - TensorBoardX lets you watch
Tensors Flow without Tensorflow
tensorbuilder (0.3.6) - A light wrapper over TensorFlow
that enables you to easily
create complex deep neural
networks using the Builder
Pattern through a functional
fluent immutable API
tensorflow-utils (0.1.0) - Classes and methods to make
using TensorFlow easier
tensorflow-transform (0.4.0) - A library for data
preprocessing with TensorFlow
tensorflow (1.5.0rc0) - TensorFlow helps the tensors
flow
tensorflow_forward_ad (0.3.3) - TensorFlow forward-mode
automatic differentiation
tensorflow_hmm (0.4.1) - Tensorflow and numpy
implementations of the HMM
viterbi and forward/backward
algorithms
tensorflow_nlp (0.0.1) - Deep Learning NLP Tasks
implemented on Tensorflow
tensorflowonspark (1.1.0) - Deep learning with TensorFlow
on Apache Spark clusters
tensorflowservingclient (0.5.1.post2) - Prebuilt tensorflow serving
client
tensorforce (0.3.4) - Reinforcement learning for
TensorFlow
tensorfunk (0.0.0) - tensorflow model converter to
create tensorflow-independent
prediction functions.
tensorfuse (0.0.1) - Common interface for Theano,
CGT, and TensorFlow
tensorgraph (3.5.8) - A high level tensorflow library
for building deep learning
models
tensorhive (0.1.1) - Lightweight computing resource
management tool for executing
distributed TensorFlow programs
tensorlm (0.3) - TensorFlow wrapper for deep
neural text generation on
character or word level with
RNNs / LSTMs
TensorMol (0.1) - TensorFlow+Molecules =
TensorMol
tensorpack (0.8.0) - Neural Network Toolbox on
TensorFlow
tensorpy (1.1.0) - Easy Image Classification with
TensorFlow!
tensorrec (0.1) - A TensorFlow recommendation
algorithm and framework in
Python.
tensorspark (1.0.6) - Tensorflow on Spark, a scalable
system for high-performance
machine learning
tensorvision (0.1.dev1) - A library to build and train
neural networks in with
TensorFlow for Computer Vision
TFANN (1.0.1) - A neural network module
containing implementations of
MLP, and CNN networks in
TensorFlow.
TFBOYS (0.0.1) - TensorFlow BOYS
tfcf (0.0.0) - A tensorflow-based recommender
system.
tfcoreml (0.1.0) - Tensorflow to Core ML converter
tfdebugger (0.1.1) - TensorFlow Debugger
tfdeploy (0.4.2) - Deploy tensorflow graphs for
fast evaluation and export to
tensorflow-less environments
running numpy.
tfgraph (0.2) - Python's Tensorflow Graph
Library
tfgraphviz (0.0.6) - A visualization tool to show a
graph like TensorFlow and
TensorBoard
…………………………………………

悲伤的我如此难以自抑，明明可以找到的，怎么却无法安装嗯？我需要自行好好找找明明是谁？ ：-）

问题分析

二话不说，直接上官网上查查看，虽然官网离我朝远隔万里，需要跋山涉水之后方可达到。翻过拿到看不见的墙之后，重要可以看到官方信息了。

官方路标如下： https://www.tensorflow.org/install/install_windows

其中所提安装步骤非常简洁，如此简洁的步骤，怎么可能出错？ 于是重新梳理了一下，难道是Python或者pip3本身的问题吗？

check pip3

pip –version

发现其为最新版本：

pip 9.0.1 from d:\program files (x86)\python\lib\site-packages (python 3.6)

那Python呢？ 官方文档中提到如下：

If one of the following versions of Python is not installed on your machine, install it now:
* Python 3.5.x 64-bit from python.org
* Python 3.6.x 64-bit from python.org

难道我安装的python是假python不成？ 估计有可能吧，难道是64bit的问题？

检查python的版本

python -v

得到了python的完整信息：

..........................................
> D:\Program Files (x86)\python\lib\__pycache__\sysconfig.cpython-36.pyc matches D:\Program Files (x86)\python\lib\sysconfig.py
>  code object from 'D:\\Program Files (x86)\\python\\lib\\__pycache__\\sysconfig.cpython-36.pyc'
import 'sysconfig' # <_frozen_importlib_external.SourceFileLoader object at 0x006A1230>
> D:\Program Files (x86)\python\lib\__pycache__\_bootlocale.cpython-36.pyc matches D:\Program Files (x86)\python\lib\_bootlocale.py
> code object from 'D:\\Program Files (x86)\\python\\lib\\__pycache__\\_bootlocale.cpython-36.pyc'
import '_locale' # <class '_frozen_importlib.BuiltinImporter'>
import '_bootlocale' # <_frozen_importlib_external.SourceFileLoader object at 0x007911D0>
> D:\Program Files (x86)\python\lib\encodings\__pycache__\gbk.cpython-36.pyc matches D:\Program Files (x86)\python\lib\encodings\gbk.py
> code object from 'D:\\Program Files (x86)\\python\\lib\\encodings\\__pycache__\\gbk.cpython-36.pyc'
import '_codecs_cn' # <class '_frozen_importlib.BuiltinImporter'>
import '_multibytecodec' # <class '_frozen_importlib.BuiltinImporter'>
import 'encodings.gbk' # <_frozen_importlib_external.SourceFileLoader object at 0x00791490>
import 'site' # <_frozen_importlib_external.SourceFileLoader object at 0x004F73D0>
Python 3.6.3 (v3.6.3:2c5fed8, Oct  3 2017, 17:26:49) [MSC v.1900 32 bit (Intel)] on win32
Type "help", "copyright", "credits" or "license" for more information.
import 'atexit' # <class '_frozen_importlib.BuiltinImporter'>

其中关于python的关键信息：

Python 3.6.3 (v3.6.3:2c5fed8, Oct 3 2017, 17:26:49) [MSC v.1900 32 bit (Intel)] on win32

“32bit” !!! 一口老血喷出，众里寻他千百度，蓦然回首bug正在这灯火阑珊处。原来是python版本的问题导致的。

修复问题

重新下载一个64bit的python版本，之后重新操作就可以了。

python -v

查看其中的关键信息：

Python 3.6.4 (v3.6.4:d48eceb, Dec 19 2017, 06:54:40) [MSC v.1900 64 bit (AMD64)] on win32

确认是64位，没有问题。

然后直接安装tensorflow:

pip3 install tensorflow

安装过程如下：

C:\windows\system32>pip3 install tensorflow
Collecting tensorflow
  Downloading tensorflow-1.4.0-cp36-cp36m-win_amd64.whl (28.3MB)
    100% |████████████████████████████████| 28.3MB 39kB/s
Collecting enum34>=1.1.6 (from tensorflow)
  Downloading enum34-1.1.6-py3-none-any.whl
Requirement already satisfied: wheel>=0.26 in d:\program files (x86)\python\lib\site-packages (from tensorflow)
Collecting protobuf>=3.3.0 (from tensorflow)
  Downloading protobuf-3.5.1-py2.py3-none-any.whl (388kB)
    100% |████████████████████████████████| 389kB 593kB/s
Collecting tensorflow-tensorboard<0.5.0,>=0.4.0rc1 (from tensorflow)
  Downloading tensorflow_tensorboard-0.4.0rc3-py3-none-any.whl (1.7MB)
    100% |████████████████████████████████| 1.7MB 182kB/s
Requirement already satisfied: six>=1.10.0 in d:\program files (x86)\python\lib\site-packages (from tensorflow)
Collecting numpy>=1.12.1 (from tensorflow)
  Downloading numpy-1.13.3-cp36-none-win_amd64.whl (13.1MB)
    100% |████████████████████████████████| 13.1MB 81kB/s
Requirement already satisfied: setuptools in d:\program files (x86)\python\lib\site-packages (from protobuf>=3.3.0->tensorflow)
Collecting html5lib==0.9999999 (from tensorflow-tensorboard<0.5.0,>=0.4.0rc1->tensorflow)
  Downloading html5lib-0.9999999.tar.gz (889kB)
    100% |████████████████████████████████| 890kB 504kB/s
Collecting bleach==1.5.0 (from tensorflow-tensorboard<0.5.0,>=0.4.0rc1->tensorflow)
  Downloading bleach-1.5.0-py2.py3-none-any.whl
Requirement already satisfied: werkzeug>=0.11.10 in d:\program files (x86)\python\lib\site-packages (from tensorflow-tensorboard<0.5.0,>=0.4.0rc1->tensorflow)
Collecting markdown>=2.6.8 (from tensorflow-tensorboard<0.5.0,>=0.4.0rc1->tensorflow)
  Downloading Markdown-2.6.11-py2.py3-none-any.whl (78kB)
    100% |████████████████████████████████| 81kB 583kB/s
Building wheels for collected packages: html5lib
  Running setup.py bdist_wheel for html5lib ... done
  Stored in directory: C:\Users\chenjunfeng1\AppData\Local\pip\Cache\wheels\6f\85\6c\56b8e1292c6214c4eb73b9dda50f53e8e977bf65989373c962
Successfully built html5lib
Installing collected packages: enum34, protobuf, html5lib, numpy, bleach, markdown, tensorflow-tensorboard, tensorflow
Successfully installed bleach-1.5.0 enum34-1.1.6 html5lib-0.9999999 markdown-2.6.11 numpy-1.13.3 protobuf-3.5.1 tensorflow-1.4.0 tensorflow-tensorboard-0.4.0rc3

然后大家就可以愉快地写代码了.

总结

问题总在认为不可能的地方发生。如果存在问题，则一定会有原因存在。见或者不见，它都在那里。