解决TensorFlow最新代码编译错误问题

朱雀 2022-05-30 05:10 178阅读 0赞

老是有个习惯，看到开源代码更新了，总是想更新到最新版，如果置之不理的话，就感觉自己懒惰了或有的不负责任了，这个也可能是一种形式的强迫症吧；

前几天晚上git pull TensorFlow，完事后也没去理它，这两天想起来还是做事有头有尾吧， 也想把学习重点转入到TensorFlow，动手玩玩才有感觉嘛^\_^。按照Image Recognition https://www.tensorflow.org/tutorials/image\_recognition先编译下，其实这个label\_image我已经编译了无数次了；但这次是想retrain image,

How to Retrain Inception's Final Layer for New Categories

不废话了，进入正题；

`bazel build tensorflow/examples/label_image/...`

但这次编译时，突然冒出一堆的错误：

`ERROR: /media/berli/DataCenter/Test/tensorflow/tensorflow/BUILD:399:12: Label '//tensorflow:tools/integration_tests/gcs_smoke_test/gcs_smoke.py' crosses boundary of subpackage 'tensorflow/tools/integration_tests/gcs_smoke_test' (perhaps you meant to put the colon here: '//tensorflow/tools/integration_tests/gcs_smoke_test:gcs_smoke.py'?)`

`ERROR: /media/berli/DataCenter/Test/tensorflow/tensorflow/BUILD:399:12: Label '//tensorflow:tools/integration_tests/gcs_smoke_test/setup.sh' crosses boundary of subpackage 'tensorflow/tools/integration_tests/gcs_smoke_test' (perhaps you meant to put the colon here: '//tensorflow/tools/integration_tests/gcs_smoke_test:setup.sh'?)`

`ERROR: /media/berli/DataCenter/Test/tensorflow/tensorflow/BUILD:399:12: Label '//tensorflow:tools/integration_tests/gcs_smoke_test/BUILD.bazel' crosses boundary of subpackage 'tensorflow/tools/integration_tests/gcs_smoke_test' (perhaps you meant to put the colon here: '//tensorflow/tools/integration_tests/gcs_smoke_test:BUILD.bazel'?)`

`ERROR: /media/berli/DataCenter/Test/tensorflow/tensorflow/BUILD:399:12: Label '//tensorflow:tools/integration_tests/gcs_smoke_test/teardown.sh' crosses boundary of subpackage 'tensorflow/tools/integration_tests/gcs_smoke_test' (perhaps you meant to put the colon here: '//tensorflow/tools/integration_tests/gcs_smoke_test:teardown.sh'?)`

`ERROR: /media/berli/DataCenter/Test/tensorflow/tensorflow/BUILD:399:12: Label '//tensorflow:tools/integration_tests/gcs_smoke_test/test_wrapper.sh' crosses boundary of subpackage 'tensorflow/tools/integration_tests/gcs`

让额有点猝不及防。。。；清楚下cache和缓存

`bazel clean;rm -rf ~/.cache`

再试试还是一样的错误！我们总是不想把时间花在这种按照调试的问题上，要用在模型，算法上，对，刀刃上，刀刃上啊！但不解决又不行，嗯，google遍查互联网，就是想找到同样错误，又能给出简明扼要的解决方式，最好是一个命令，别他们一堆一堆的邮件，BB的不知道啥内容，说实话，浪费时间，也没心情陪你们玩。

**苦心人终不负，百二秦关終属楚！最好终于找到原因：就是google最新的TensorFlow代码已经不兼容低版本bazel编译器了！！可我的bazel是0.8.0，也不低啊**

`sudo apt-get update ;sudo apt-get upgrade bazel`

升级后我们检查bazel version，下面命令

`bazel version`

`Build label: 0.11.0`

`Build target: bazel-out/k8-fastbuild/bin/src/main/java/com/google/devtools/build/lib/bazel/BazelServer_deploy.jar`

`Build time: Thu Nov 9 04:43:10 +50056 (1517474666590)`

`Build timestamp: 1517474666590Build timestamp as int: 1517474666590`

嗯 ，继续编译lable\_image`bazel build tensorflow/examples/label_image/...`