Support for build on hosted on (ana)conda

[ewengillies]

I see that some of the continuous integration scripts support conda builds, although not all the dependencies are installed this way. Is there any hope of seeing a build on conda soon for Linux x86_64 systems?

The reason I ask is that I have accounts on numerous batch systems, none of which I have root access or have any way to use docker. They're all linux-based though, as is the norm. So far as I know, this is the case for many researchers.

It'd be great to see a way to quickly install REP on these systems. This would:

• Cut down on the time needed to introduce people to REP
• Hook into the environment management and environment logging provided by conda
• Easily and quickly deploy REP on supercomputing nodes while requiring little of their filesystem

This is especially useful for ensuring the ROOT install is sane. I know there has already been a lot of work in the direction of making REP easy to access and install. Perhaps this could be a healthy addition?

[arogozhnikov]

Well, I don't think we have enough time to support and test conda builds, but you can use conda right now as it is done in integration scripts.

This isn't much harder:

• download (wget) appropriate .yaml for environment
• e.g. for python 2:
  wget https://raw.githubusercontent.com/yandex/rep/master/ci/environment-rep2.yaml
• conda env create -q --file environment-rep2.yaml
• conda activate rep_py2
• pip install rep

This should work (at least after we update package in pip)

[arogozhnikov]

@e-gillies-ix did you try the instructions above?

[ewengillies]

Yeah, this recepie works fine. I guess I was hoping to have it all in the .yaml file, i.e. excluding the last pip install rep command, but this really isn't worth the extra effort...

Thanks! Ewen

[anaderi]

@e-gillies-ix , you can add rep to environment.yaml like this:

dependencies:
  - pip:
    - rep

[ewengillies]

ahh, okay, this is a bit cleaner then. Thanks!