Hi there! (First of all, a public thank-you to Steve for pinging me
off-list and directing me to this discussion.)
First of all, I must say it was a happy surprise to see the Kepler
name suggested and the positive reaction that it got on this list. If
anything, it means our efforts back in the day still leave a positive
residual perception. :) So yes, I'm all for the idea. Still, some
interesting questions were raised in this thread and I'll try to cover
them to the best of my knowledge while bringing some context:
A quick recap on Kepler: Kepler was envisioned by André Carregal as a
research project to make Lua a viable language for web development, in
the sense that things like PHP and Ruby are nowadays viable, ie,
having the tools so that a coder can concentrate on their app rather
than reinventing wheels. So yeah, that goal included a good deal of
the "batteries" as well as figuring out distribution issues. A major
part of this project, also, was to bring together people who were
working seperately in pieces of this puzzle, including the politics to
cooperate with pre-Kepler projects such as CGILua, LuaSocket and
LuaBinaries. This also included getting funding -- Carregal spent a
lot of time working to get government grants for the research project
(which was officially a project involving PUC-Rio and the company he
worked on as well). People like Fabio and myself (both grad students
under Roberto at the time) were funded by this project to work on Lua
modules. LuaRocks was created because of Kepler, and I was paid to
work on it for two years. LuaForge was also created and its
maintenance funded by Kepler.
Eventually, we didn't get a grant renewal, plus the overall landscape
wasn't much in Kepler's favor (it never took off as a killer app for
Lua), and the project mostly died off. Some of us "inherited" the
modules we worked on and remained on them as volunteer side-projects,
working on our free time. I kept LuaRocks, Fabio kept Orbit and
others, and so on. Note that this includes stuff that I perceive as
key infrastructure such as LuaFileSystem. Fabio takes care of its
github repo, and released version 1.6.2 about 6 months ago. André
Carregal, on the other hand, effectively left the project (and I
really can't blame him for the burnout, after all the effort he went
through -- he's an unsung hero of the Lua ecosystem (or perhaps
"sung", since Yuri's book, which I heartily recommend for insights on
the story)). The project was reshaped not as a platform that made
"releases" but as a general hub for all those subprojects (the
availability of LuaRocks helped to diminished the reliance on one big
tarball).
The LuaForge shutdown was probably the major effect of the project's
twilight. Carregal tried as best a he could to keep the service going,
but without funding to hire someone to babysit GForge (and the growing
expectations of the Lua userbase), keeping it up with regular
shortages wasn't an option. He tried handing over LuaForge to "the
community", but as most of us remember, lots and lots of discussion
went on, but nothing as concrete as the old LuaForge ever emerged.
Now that we're done with the flashback, we get to where we are now.
A few thoughts on the matter:
* It looks like the effort you guys are organizing is currently a lot
more focused than the LuaForge 2 brainstorm we went through. I guess
it's because we're starting from something concrete (LuaForWindows,
LuaDist) and building from there. Still, for things such as a website
presence, documentation, etc, there's still a lot to be done. Focus is
key so it doesn't go the way of LuaForge 2.
* As for the use of the name, I'm not sure on its legal status, but my
guess is that Carregal, Fabio and anyone else who was involved in
Kepler would be happy to hand it over to a project that thrived to
ship a Lua platform. The intention to hand over the LuaForge name was
there before. It ended up not happening (
luaforge.net still points to
the backup site Yuri (and Fabio?) put together) because there was
never a "version 2" website done which could host that content.
* As for the logo, we never had branding problems with Kepler before.
LHF's comment on the Lua stylesheet surprised me a bit, to be honest,
but we never used material straight from the
lua.org website. Roberto
himself used the Kepler style on the LPeg website (IIRC, advised by
Carregal even). I think Kepler (the old or the new one) should keep an
image close to that of Lua (dark blue, sans-serif fonts, etc). The
logo is close to the Lua logo but not the same (gray outline,
different font). Lots of projects use variants the Lua logo
(LuaSocket, LuaRocks, etc) and the Lua authors never complained.
* On naming opinions in general: I think names such as LuaForge,
LuaForWindows, Kepler, are "strong brands". LuaForge due to its mental
association to RubyForge, SourceForge: it's an obvious name for a
catalog of projects. To be more modern, perhaps LuaHub would be the
other option, but
luahub.com is already in use. LuaForWindows is
another obvious name, and the project got support even from
lua.org
(heck, it's got a direct link from the Lua downloads page, and it's
been there longer than LuaRocks ;-D ). I'd like to think Kepler is a
strong brand as well after all the effort we put into it, and your
reaction here seems to agree. I think there's room for all these names
(as well as LuaDist, LuaRocks and others) as we're talking about
several different things here.
> My assumption was that Kepler was linked directly to PUC/Rio and that
> their stylesheets, Logo and overall branding was directly linked to
> the core Lua team's.
* It was linked but not that directly. The Lua team always wanted to
keep clear that the crazy "platform" stuff we were doing was not
coming from them (even though we occasionally interacted/consulted
with them and even contributed some modules, especially in pre-Kepler
days (now Roberto maintains basically lpeg only and LHF has his own
collection of modules)). Since 5.something they've been using Lua.org
as a copyright, to further separate themselves from Tecgraf, PUC-Rio
and avoid confusions of any kind. So, there shouldn't be worries about
that. (But I'm not a lawyer, blah blah... anyway, I can always ask
them. Roberto is currently my PhD advisor, after all).
> I am not sure. The kepler list[1] is active looking over some of the topics
> I must conclude that may of them overlap with the intended purpose of
> "batteries" project. Best to ask there.
Yes, there is overlap in the stated goals, as there is some overlap
between LuaDist and LuaRocks, for example. But nobody from the
original Kepler team is working on a "batteries"/"platform" package
anymore. That doesn't mean we wouldn't love to see it happen.
> I would also make Xavante, Orbit, and and other Web technology a base
> requirement for the install. This would keep in line with Kepler's original
> goal.
That would be nice, but I don't know how much that's necessary
(especially given all this things are one command away, with LuaDist
and LuaRocks). Some people will remember that Kepler has always been a
web-oriented platform for Lua, but I wouldn't mind seeing it as simply
(and more generally) a platform for Lua-based development (as opposed
to using Lua in a C-based development). That would be my suggestion
for the scope: the embedded/games people will always use Lua
differently and will have no use for
LuaForWindows/LuaDist/LuaRocks/Kepler. They're a huge part of the Lua
community, but there's another part which wants to work with Lua as a
base (ie, with a Lua "main function"). Those are the folks who need
the batteries.
Since I'm writing a long message anyway, I'll chime in some other
comments from my catching-up with the list:
> However, I feel we can have multiple persons responsible for various
> parts of the projects. Eg. Steve D. would be responsible for module
> policies (what tools to use, how to structure them etc). Ryan for module
> evaluation and selection (what is in, what is out). I am willing to be the
> person responsible for the technical parts such as CI, tools and of
> course maintainer of the more complex builds that sometimes need
> to be done.
Fully agreed. This is the kind of "focus" I was talking about. Too
much decision by committee will lead to a "LuaForge 2" situation. I'm
all for having gatekeepers (benevolent sub-dictators) for these
concerns (and I agree with the above nominations too).
> This boils down to the difference in philosophy. LuaRocks is hands off,
> it works for me approach. LuaDist goes the hard way and maintains
> everything.
Agreed, and both have pros and cons which make them suitable for
different parts of the project. The LuaRocks approach is inviting to
developers (and we know many module authors would shun at CMake (I
spoke to a few)) and scales (once we automate things further with the
MoonRocks project or the like) -- look at other language's repos, with
thousands and thousands of modules: the potential is there. The
stricter approach of LuaDist is perfect for maintaining the core
platform, ensuring it is portable, etc. We don't need this kind of
quality assurance for every single module on Earth (nor we have the
manpower to do it), but at least having them cataloged a rock is
better than nothing (see the fast-growing rate of ruby and npm repos).
The overlap between LR and LD will probably cause confusion to
newcomers, though. It's something to be sorted out.
I see a lot of potential for this effort, and I'm willing to help any
way I can (which unfortunately is not a lot lately). But count on me
to talk to Kepler/PUC people, etc.
-- Hisham