lordgamez commented on a change in pull request #1047:
URL: https://github.com/apache/nifi-minifi-cpp/pull/1047#discussion_r620439674
##########
File path: .github/workflows/ci.yml
##########
@@ -215,8 +215,14 @@ jobs:
sudo ln -s /usr/lib/x86_64-linux-gnu/odbc/libsqlite3odbc.so
/usr/lib/x86_64-linux-gnu/libsqlite3odbc.so
echo "PATH=/usr/lib/ccache:$PATH" >> $GITHUB_ENV
echo -e "127.0.0.1\t$HOSTNAME" | sudo tee -a /etc/hosts > /dev/null
+ docker pull lordgamez/tensorflow-cc-binaries:ubuntu-20.04
Review comment:
I'm not familiar with tensorflow either, but as I checked the APIs and
the distribution it seemed that the only way to use the C++ API is to build
your own libraries from the tensorflow sources. There is an official docker
image and pip packages, but those only provide python packages. Additionally we
currently use the [tensorflow_cc](https://github.com/FloopCZ/tensorflow_cc)
wrapper library instead of the official tensorflow libs in our extension, which
also includes only the source files in its releases. For me it seemed the the
most viable option to package these libraries in a docker image, this way we
have a documentation of the builds process as well in the form of the
Dockerfile. But I'm open for any additional ideas, if we have some better
options.
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
