What is the best way to quickly update PyFR to the latest version without losing any dependencies like MÉTIS for example. I do not want to go through the hassle of libmetis to gklib and all of the other setup steps. Is there anyway to quickly upgrade PyFR to avoid having to do everything from scratch?
Typically, all of your python dependencies (gimmik, h5py, mako, etc.) should be installed in a saved python virtual environment. Some of the backends have additional dependencies (e.g. libxsmm, metis, etc.) and those typically only need to be compiled once on your platform unless requirements change due to updates to the code, and then environment variables tell PyFR where it can find those libraries.
For me, updating PyFR is simply then a matter of either fetching the latest changes and pulling those changes to my local copy of the develop branch, or alternatively checking out the latest release branch locally, and switching to it. Then it’s just a matter of activating the python virtual environment and running PyFR.
Hopefully that adds some clarity on updating PyFR…
2 Likes