Unable to disable build of AWS, HDFS, Kafka and GCP
lissyx opened this issue · 4 comments
System information
- Have I written custom code (as opposed to using a stock example script provided in TensorFlow): no
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): any
- Mobile device (e.g. iPhone 8, Pixel 2, Samsung Galaxy) if the issue happens on mobile device: any
- TensorFlow installed from (source or binary): source
- TensorFlow version (use command below): master
- Python version: N/A
- Bazel version (if compiling from source): N/A
- GCC/Compiler version (if compiling from source): N/A
- CUDA/cuDNN version: N/A
- GPU model and memory: N/A
- Exact command to reproduce: ./configure
Describe the problem
When building TensorFlow as a library (and in our case using --config=monolithic
), we used to remove as much as possible of the not used features. Recently, this was removed from master in d56c298 for AWS, HDFS, Kafka and GCP.
I understand that that having to deal with multiple optionnally-disabled features can be a huge burden for the future, but is it possible to have some way around ?
Discussed offline.
I will need to reintroduce a way to disable building these into our build system.
I am thinking having these default on, but having --config
options such as noaws
, nogcp
, nohdfs
and nokafka
What do you think?
I think that any solution that is okay for you in term of complexity and allows us to disable should be good :)
Thanks for that!
@gunan Hello, I'm testing that on current master, my build command line does include --config=noaws --config=nogcp --config=nohdfs --config=nokafka --config=noignite
and yet, checking out the bazel-out/
subdirectories, I find .o
and .d
files that have been rebuilt, under aws-cpp-sdk-core
, aws-cpp-sdk-kinesis
and aws-cpp-sdk-s3
, am I missing something ?