[CMake] CMake with Lua Experiment

[Brandon Van Every]

On Nov 26, 2007 6:03 PM, Pau Garcia i Quiles <pgquiles at elpauer.org> wrote:
>
> Before Lua is added as a language to CMake, I'd like to ask a
> question: is there really a need for Lua (or any other language) or is
> the current CMake scripting Turing-complete and able to do anything
> which could be done with Lua?

That's a false dichotomy.  Turing Completeness doesn't say anything
about whether a language is a PITA.  To me there are 2 questions.  (1)
is there anything seriously objectionable about CMake script as it
stands now?  (2) is it better to improve CMake script itself, or to
embed a 2nd language that doesn't have the problem?

Many of us agree that lack of scope for variables is a problem.
However, when we last spoke, acceptable solutions seemed to be
possible within CMake itself.  A question is, how long will they take,
vs. embedding a language which doesn't have a scoping problem?

I think CMake's regular expressions are quite a bit less than ideal.
Adding the Perl Compatible Regular Expressions library to CMake is
certainly possible.  It would be a certain amount of work, and I've
wondered if the work would be worth it to me personally.  If CMake
adopts Lua for other reasons, I can get Posix and PCRE regexes for
free, as people have already implemented wrappers for them.  Lua's
core syntax has better regexes than CMake in some respects, but it's
missing the branch "|" operator.  I'm not sure whether I'd find Lua's
regexes adequate in practice; it would be a contest between the
"better" and the missing features.

CMake's documentation, as mentioned, is objectionable.  It's getting
better, but we need definitive ways for people to learn quoting and
escape behavior, list vs. string types, etc.  "Chapter oriented"
material.  Documentation in CVS is getting better, a new version of
"Mastering CMake" will be available in early 2008, and we're perfectly
capable of revamping the wiki if we have the energy.  A new language
such as Lua, however, has a lot of documentation "for free" that we'll
never have to maintain.

Lua is a sufficiently popular language that it can attract people to
CMake simply because Lua is being used.  It is perceived as a "fast"
scripting language.  When I think about converting people and gaining
the "ultimate masses" of CMake users that we will one day have, I
think Lua has a lot more strategic marketing power than CMake's Yet
Another Scripting Language.  Just as experienced CMake users are
currently objecting to having to deal with another language, the rest
of the world objects to having to learn CMake script.  And at present,
I think there are a lot more techies who don't use CMake than do use
CMake.

At present I'm open to the possibility of Lua.  I don't have to have
it.  Everything on my list of objections could be solved by improving
CMake script.  I think there are valid reasons to investigate Lua.


Cheers,
Brandon Van Every