If it is something that can be initialized once, I often do it at the start of the program as you mentioned. Basically, the entry point to the program can do whatever it needs (including side effects) to setup the program.
I am currently working on an article for how we are managing side effects interleaved with logical decisions. Here is an extended example that I plan to explain in the article. It is based on the MVU pattern. And in fact, for the front-end we also augment the MVU pattern with a Perform function to run side effects. Example here.
For randomness, I have an old article that addresses how to shuffle without side effects.
I also highly recommend Mark Seemann's blog for articles on functional architecture. Here is a good one, for example.
If it is something that can be initialized once, I often do it at the start of the program as you mentioned. Basically, the entry point to the program can do whatever it needs (including side effects) to setup the program.
I am currently working on an article for how we are managing side effects interleaved with logical decisions. Here is an extended example that I plan to explain in the article. It is based on the MVU pattern. And in fact, for the front-end we also augment the MVU pattern with a Perform function to run side effects. Example here.
For randomness, I have an old article that addresses how to shuffle without side effects.
I also highly recommend Mark Seemann's blog for articles on functional architecture. Here is a good one, for example.
Thank you for this and thanks for those links. I'll definitely check them out.
Love the shuffle idea.