Hey folks, like @kaylangan I'm a PM on Azure Pipelines. I know woefully little about C++, having not touched it since undergrad, so bear with some dumb questions. Will you see speedups from clcache within a single build job, or does it only help if you can cache across runs? We don't offer any persistent storage between runs in the hosted pools.
We can definitely investigate having precompiled Boost available. I've opened a feature request to have a C++ expert investigate: https://github.com/Microsoft/azure-pipelines-image-generation/issues/263 Performance characteristics of C: vs D: using Azure VMs the way we do is... complicated :) If you're finding that for your workloads, the D: drive is faster, then you should use it. (That's likewise why we run the agent itself off the D: drive.) We don't make D: the default for a lot of things due to space constraints. [ Full content available at: https://github.com/apache/arrow/pull/2674 ] This message was relayed via gitbox.apache.org for [email protected]
