bmmalone / urlearning-cpp

This c++ project implements a number of algorithms for learning Bayesian network structures using state space search techniques.
MIT License
8 stars 4 forks source link

What development environment to use? #1

Closed bmmalone closed 7 years ago

bmmalone commented 7 years ago

This project uses boost rather extensively. Due to this problem, eclipse may not be the best IDE for this project. On the other hand, there could be some work around to address the indexing problem with boost/eclipse.

Building with Bazel ensures we don't need much on the path, but it would still be nice to have something reasonably light-weight to keep setup/configuration on new systems minimal.

bmmalone commented 7 years ago

Compiling the boost index requires changing some parameters of the indexer; namely, it seems necessary to increase the RAM available to java ("-Xms1G" and "-Xmx3G" in eclipse.ini) and to set the "Use a fixed build configuration" for the indexer. After this, building the index takes about 2 hours on an otherwise-idle machine.

Since the boost indexing seems resolved for the time-being, we will use Eclipse moving forward.