Thanks for the stream! From what I understood, a virtual environment created in conda with all the project dependencies installed in it can be exported to a single yml file called environment.yml. I wonder if this file can be distributed and anyone who wants to recreate the same virtual environment as the original developer, can do so. This will relieve others from having to even make use of requirements.txt or install dependencies one by one. But, I missed to notice if the size of the file is big and whether this approach of distributing your software scales well?
What do you use inside a conda virtual environment to install packages - conda install
or pip install
?, and do you have a suggestion or an experience to share? From what I've read in the Anaconda documentation, conda install
is the reccomended approach but then it prevents you from generating a requirements.txt file in a pip compatible format. Although you can still share the environmet.yml so that conda users can replicate your environment, I believe in this case non-conda users are those to whom you will not be able to distribute your software. Would like to clarify this and find out if I have misunderstood anything here.
Thank you for a very nice stream! It was very clear, easy to follow, and had a relaxed and cozy vibe. From the spectator side, it really felt like all three of you got a chance to speak and were naturally included to bring the discussion forward (something that even veteran hosts fail with frequently). Here are some feedback on minor things that could maybe improve the stream: (+1)
Thanks for your feedback!