No 
Thanks, I will try them as soon as possible.
To return to the original topic, it seems most people either don’t care much (if at all) about what GitHub is, or they believe adding a restriction about it in the license makes the software less “Free”. I still don’t see why such a restriction is that serious, given the plethora of similar sites the user is free to upload the code.
However, the last thing I want is to have a “my software my rules” attitude. Furthermore, I guess I have wasted too much of my time (and yours) on this already, and it’s time to move on. Therefore, I decided to just add a line in the end of the zlib license text saying
Although not required by the license, please do not upload this software on GitHub.
This makes it clear it is not a restriction, just a request.
As for why zlib… I don’t know, it sounds a little better to me, but BSD-3-clause or MIT are great choices as well. For larger projects I would probably go for LGPL.
Thank you all for your time.
I understand why you did that but, in my humble opinion, anyone using SymPy (or any other library, especially of that magnitude) must cite your work, even if it is not a license requirement. If they don’t do so, that’s completely unethical, therefore completely unacceptable. There is no other way to put it.
It doesn’t matter if it is technically allowed. Unwritten ethical laws should be considered much stronger than just a license.
I agree. People cite the SymPy paper, it has over 1,200 citations. I think the system works. The software is fully open source, no other strings legally attached. And our “wish” to get cited, and people fulfill it.
So that’s why I recommended it to you. If you write your wish in the README, people will generally obey it, that has been my experience.
2 Likes
That’s just a question of principle about the definitions of what is free software. But feel free to add your own restrictions. Ultimately what is important is that you are free. Human beings freedom is more important than software freedom, isn’t it?
1 Like
You may also consider the new Codeberg forge (cited above). A few Fortran projects are already living there:
That’s correct, no doubt about it. However adding such a restriction while basically none else does has a risk to sound too demanding. And that usually has the opposite result than the one desired. That’s exactly why I started this thread, to see what people think about it - and if they have seen something similar in the past.
Maybe just a recommendation will have more impact than a strict rule.
To be clear, I am not opposed to GitHub just because I dislike its owner. I won’t deny that my opinion about the company behind GitHub, based on countless of facts for decades, is the worst possible - to put it politely. That alone would be enough to avoid GitHub. But even if this wasn’t an issue, there are two reasons, irrelevant to the owner, that I think they are serious problems and they can’t be ignored.
- Centralization: The overwhelming monopoly that GitHub has as a software host is a problem that most people choose to ignore. Do I even need to explain why any monopoly is bad? Except the board game, that is (who didn’t play Monopoly as a kid?) - and I’m not quite sure about that exception either.
- Copilot: I’m not going to discuss the problem with AI itself here, how often it fails generating decent code, and the more general question if AI in programming or elsewhere is a good thing. The problem is Copilot collects data from the vast amount of code hosted in GitHub. This can lead to legal issues GitHub just ignores. @HugoMVale already mentioned a website discussing the matter, but it’s pretty easy to find others as well. I do not want to contribute a single line of code to Copilot. And I won’t do it myself, but if someone uploads my code on GitHub then I do contribute to Copilot, without my consent.
There are other issues with GitHub, but those two should be more than enough. But in a world where data collecting and selling becomes the norm and most people don’t give a dime about that, or they even help it voluntarily in “social” media, I start feeling… old fashioned.
1 Like
Any code written in Fortran 77 or older is obfuscated enough beyond understanding 
1 Like
With an (in)appropriate amount of OOP you can also obfuscate anything you like
2 Likes
100% agree on that.