I am not a layer, but here is what I proposed and was validated in my jurisdiction:
- Include ALL license files from any included MIT code, be it forked, copied as-is, etc.
- If the license files are copied, or renamed (to avoid overwrites), you can name them LICENSE- (e.g. LICENSE-numpy).
- Create a main LICENSE that is your license, and at the end of the file you list all the MIT software along with their copyright holders, and the relative path to the related license file.
Flux Capacitor 1.0
Copyright (c) 2020 ACME Corporation, all rights reserved. Yada, yada, yada...
Includes other software related under the MIT license:
- Soft1, Copyright 2013 John Doe. For licensing see /LICENSE/LICENSE-soft1.txt
- Soft2, Copyright 2010 Mary Smith. For licensing see /LICENSE/LICENSE-soft2.txt
- Soft3, Copyright 2010 Dr Elwood. For licensing see /SOFT3/LICENSE.txt
Ideally, you would leave the license files in each software/package directory with the original file name (as in the example of
Soft3 above), but that is not always possible, particularly when merging a lot of code form different authors into a whole new product. For
Soft2, I moved the license files to /LICENSE directory instead. These are just examples and in real life I would strive to keep a convention across all license files.
According to my legal council, it is perfectly valid to rename or move the license files, if this was done in order to provide further clarity on what is what.
Consider for example, software that isn't redistributed, but has some sort of about box. They could have a section that says:
Portions of this software are copyright of their respective authors and released
under the MIT license:
- Soft 1, Copyright 2013 John Doe
- Soft 2, Copyright 2010 Mary Smith