Re: What counts as a fork?

[Paige DePol <lual@...>]

steve donovan <steve.j.donovan@gmail.com> wrote:

> On Wed, Nov 29, 2017 at 12:46 PM, Paige DePol <lual@serfnet.org> wrote:
>> things about a living language. My only point in all of this was that any
>> version of Lua created by anyone who isn't one of the original developers
>> is, by every definition I can find, creating a fork of Lua.
> 
> Although not necessarily in an interesting sense.
> 
> Doing a Github fork is obviously just a development convenience -
> there is still an upstream and a desire to push things upstream.
> 
> Re-implementations of Lua exist - they remain Lua as long as they
> preserve syntax and semantics.
> 
> It would be wildly exciting (in a bad way) if one of those independent
> projects starting to claim that they were the new Lua, and try
> mobilize community around that.

Yes, GitHub forks are often (even mostly[1]) a development convenience,
I will agree on that point. Though, by the definition of "fork" it still
is creating a copy for independent development. So the usage of the term
is still valid, even if in the practical sense it may not necessarily have
the same meaning politically as it used to.

I was never trying to address the political sense of the word "fork",
just the literal dictionary definition. Honestly, I never really even
considered the political aspect of the word as it is not something I
have a lot of experience with. To me, "forking" has just been something
you do when you want to develop code that wasn't yours to start with,
and not just because of Git, but since I started using source control.

As for someone claiming to be the "new Lua", that would be a bad fork in
the political sense then. I am going to guess that situation has probably
occurred with other projects before.

What if another variant of Lua came along that was both new, but it also
preserves backwards compatibility? So, a new language with new semantics
and features, but also still able to load and run vanilla Lua source?

~Paige

[1] According to a study ~70% of GitHub is duplicated code!
    https://www.theregister.co.uk/2017/11/21/github_duplicate_code/