HomeBrew and Python installation on Mac Book PRO

Because you are installing third-party Python libraries, you do not want to install these with the Apple-provided Python versions. Even if your school's instructions suggest that this is okay, don't do it. I use the plural here because, in addition to Python 2.7.16, Xcode also installs a 3.8.2 version in /usr/bin/python3. Leave these alone.

It is important that you know in advance what versions of Python that those school recommended Python libraries support because if they want Python 3.8 (example), they will likely break with a newer version. The link in the next paragraph gets you to recent versions of Python3 for the Mac.

What I have personally done is visit Python.org and download their Mac-friendly Python 3.9.5 installer. It installs its library in /Library/Frameworks and then soft links from its library, the binaries back into /usr/local/bin. Simple as pie, and nothing Apple-related gets stepped on.

Once you have the preceding paragraph accomplished, you should adjust your PATH in your ~/.zshrc, or ~/.bash_profile shell dotfiles:

export PATH=".:/usr/local/bin:${PATH}"

Homebrew, on Intel Big Sur, still uses the /usr/local tree for its package installation, and /opt/homebrew tree on M1 Macs. If from above, you already have installed some version from Python.org, although any brew package dependencies that require it to install (currently) Python 3.9.5, it will build them, but not attempt to move those Python binaries from /usr/local/homebrew into /usr/local/bin. No big deal.

When I intend to install third-party Python libraries, I create a Python virtual environment to isolate that content from the original Python library by stuffing it all in an arbitrarily named folder, the following example is in my home directory:

/usr/local/bin/python3 -m venv ~/py39venv
source ~/py39venv/bin/activate && cd ~/py39venv
# install your libraries in this venv
# and when done
deactivate && cd -

Once I have changed the directory to ~/py39venv, I am free to install third-party libraries in that folder, and under no risk of screwing up my Mac.

On your Intel Mac, homebrew expects that /usr/local exists before you install homebrew. Once brew is installed, you can

brew update
brew upgrade
brew cleanup

It is a good idea to see what your package dependency tree looks like for an uninstalled package before you pull the trigger:

brew deps --tree --installed package_name
# get information on it
brew info package_name
# then you can install it
brew install package_name
# list your installed packages
brew list
# and then cleanup again
brew cleanup



mauede33 wrote:

The school has published the installation instructions on the following website:

https://github.com/SyneRBI/SIRF/wiki/SIRF-SuperBuild-on-MacOS#Install_XCode_and_Homebrew

but they are quite confusing to me.

They are also years out of date - and, by their own admission, untested.

In fact, XCode is already installed on my Mac. I still need HomeBrew and the latest Python.

Just so you know, HomeBrew is for people too lazy to properly build software. Then they dump people like you on these forums so that we can sort out their mess for them.

There are instructions about integrating SIRF and CIL with MathLab, which I would like to use. Unfortunately, school instructors will demonstrate SIRF and CIL usage with Python.

My concern is to screw up my currently working Mac configuration.

That is a valid concern.

My recommendation would be to download Parallels Desktop from the Mac App Store: https://apps.apple.com/us/app/parallels-desktop/id1085114709?mt=12

Do NOT be tempted or swayed by marketing to install the other version of Parallels Desktop that isn't in the Mac App Store. You don't want that one. You want the one in the Mac App Store. It will cost $100/year, but it will be well worth it.

Then, install the free Ubuntu Linux operating system in Parallels. This is one of the built-in options and very easy to do. Then, you can follow these instructions: https://github.com/SyneRBI/SIRF/wiki/SIRF-SuperBuild-Ubuntu

If you do this, you don't have to worry about Python 2.7 vs Python 3.x. You don't have to worry about HomeBrew. You don't have to worry about permission on /usr/local. You don't have to worry about any of the other random things are are likely to fail during the Mac install process.

If your only goal is to install Python3, and the other libraries that the school recommends can be installed by Python3's own package installer (pip), or a means other than homebrew, then you really don't need homebrew at all. Keep it simple. Use the appropriate Python3 installer from Python.org, and the keep your Python development environment contained within a designated venv folder.

Or follow etresoft's advice.

mauede33 wrote:

Parallel Desktops is like a virtual machine.

It is not like a virtual machine, it is a virtual machine.

The image reconstruction library that is the object of the school will benefit from the availability of GPUs and/or multi-core platform.

Will whatever runs in the virtual machine have access to all the cores installed on the Mac? Will it have access to multi-threading?

You can configure it to have access to all the cores. GPUs are always challenging with a VM.

SIRF and CIL are a framework for image reconstruction from CT, PET MRI scans. I suspect they use iterative algorithms that are number-cruncher processes. I am concerned about the speed.

Don’t be concerned about speed. You won’t have much of that.

My recommendation is based on getting you up to speed with an environment that you can use to follow along in class. It will be grossly unsuitable for any real work. You are welcome to try to get it running natively in your Mac instead. You might want to put the medical classes on hiatus for a few years while you get that computer science degree. You will also need a few years learning how to build and install poorly-written open-source software on a Mac. Or you can take a chance with Homebrew. Who knows? It might work.

However, even if you get it running on your Mac, all you would get from that is slightly higher performance while running the demo data sets for school. If you want to do any real work or make use of this experience to get a job, you will have to go back to square one and get it all working on Linux. I’m suggesting you cut out the middle-msn and just learn to do it on Linux. Then, when you want to do a bigger project, you could run it on much more powerful machines on AWS. Telling a potential employer that you got it running on your Mac with Homebrew isn’t going to impress anyone. Telling them you got it working on AWS will get you a much different reaction.