2017-04-20 18:37 GMT+02:00 Andrew Starks <andrew@starksfam.org>:
> My understanding is that the authors of Lua probably wouldn't mind if
> Lua was adopted for more of the roles that _javascript_ and Python get
> used for. I also understand that they do not want to manage and more
> importantly, the do not want to organize that effort. They don't even
> want to bless it.
>
> The "standard library" often comes up in various threads. Progress as
> a general purpose language means having an ecosystem and that means
> picking some basic libraries, enforcing one way to version things, one
> package system, etc. Without some way to enforce limitations new users
> in this segment are left befuddled with too many plausible options
> that have very little commonality and they have little sense of what
> is current and what is abandonware. Apart from LuaRocks, it's
> difficult to know which horse to back.
>
> LuaRocks is an extremely successful project, but a project does not
> enforce much of anything, including quality standards. There have been
> other excellent efforts, but they haven't included a political
> structure, either.
The Internet is a democratic anarchy. If some hardnosed individual starts
up a website of "Lua modules that I use", with good documentation and
fail-free builds over the platforms that Lua itself supports, people will
vote with their feet.
The trouble with efforts along these lines in the past (not wishing to
offend, I won't name them) is that they have been too permissive.
Another JSON/YAML/XML parser? Welcome on board, my boy,
there is always room for one more. Not quite a drop-in replacement
for the others? Never mind, heh-heh, they can read, can't they?
Speaking as someone who recently needed modules for curl and
json, I can say that it was a PITA to choose even between the ones
on LuaRocks, let alone the private repositories kindly offered by the
very helpful people who replied to my posts (and those replies are
what I treasure about Lua-L, don't misunderstand me).
I agree with Andrew — I would dearly love a selection with just one
trustworthy instance of each, like Python provides.
I don't think behemoth integrated multi-modules with substantial
intradependence (again, I'm not naming them) are the answer.
We need good independent modules of managable size, tested
and vouched for by someone with gravitas.
Don't try to get consensus on Lua-L. Or even a vocal minority in favour.
Just get on, do it (properly), offer it, and if it is good enough, it will
become the de facto standard.
After all, the Lua team did not ask the world whether it wanted Lua.
They designed it to please themselves and made it available.
Yes. I was internally mindful of not just posting a lot of arm waving. I'm puzzling it out, but I think that a foundation with a simple mission and polite governance might be the best vehicle towards getting it done. I'm not the doer guy and I'm not qualified to make informed opinions about important technical considerations (or even what those should be).
But I can push and I can organize and I can try to get it started. I'd like to get a sense for the appetite, especially because it will require time in meetings and real work.
-Andrew
-- Dirk
