Error during parseyapi.sh execution: Failed to load Skylark extension

[skerit]

I can't install the image, it fails with this error:

ERROR: Failed to load Skylark extension '//tensorflow/tensorflow:workspace.bzl'.
It usually happens when the repository is not defined prior to being used.
Maybe repository '' was defined later in your WORKSPACE file?
ERROR: cycles detected during target parsing.

[johndpope]

ahhh - so fragile - I spent weeks on this - it did actually work - end to end. I think try getting it to build locally. It looks cpu specific I'm afraid.

[skerit]

I think I found the main issue.

It seems like you used a hash commit for every git repository, so the problems not there, but for the syntaxnet docker file you're basing it upon you use:

gcr.io/tensorflow/tensorflow:latest-devel

I switched it to 0.12.0-devel.

Unfortunately, then I bumped into a problem with zlib. Apparently, syntaxnet uses a specific zlib package from the main zlib.net site, but once zlib releases a new version it removes the previous one :/

That was easily fixed by chaning the syntaxnet workspace file, though.

[skerit]

I also had a problem with this line:

find / -name "parsey_api" -size +512k -exec cp -t "{}" /work/serving \;

In my case, the -t flag expects the source to be a directory, but that's obviously not the case. Removing the -t flag fixed that issue.

[johndpope]

frankly - I always wanted the binary file parsey_api to be simply downloaded from google drive as part of docker initialization rather than going through the pain of building it. just need to throw it up in the cloud and rejig the gist script.

[qysnn]

Downgrading Bazel from current version to 0.3.2 solves this problem for me. As for the zlib problem, https://github.com/tensorflow/tensorflow/issues/6594 gives servaral ways to fix it.