Re: Propsoal: a lua dialect without nil

[Axel Kittenberger <axkibe@...>]

> Part of it is the  'Why can't Lua be like X' theme, where X is some
> language du jour.  Some of it comes from the old idea that humans
> cannot be trusted, which is very strong in the design of Java. (And
> Java is considered pretty lax by the real strong-typing enthusiasts)

I feel innocent on this "be like X", cause I didnt know of any X with
that nilless-traits.

> Such discussions can be intellectually stimulating or plain
> bikeshedding (like the old men in the pub) depending on the maturity
> of the participants ;)  They help refine our thinking.

Indeed, and contrary to Luiz and Roberto noone else gave his/her
statement to read every message on the list. If you feel pestered
with, just skip it please

> My personal roadmap is going to shift more towards environments and
> editors, the tooling aspect of the ecosystem which is so important for
> productivity with larger applications; David M's LuaInspect is a very
> promising way forward, pushing static analysis beyond what we might
> naively think is possible with a dynamic language.

Indeed, and to shed a story that leds way away from Lua. Years back,
within the static/compiling paradigm, I wondered why we stick to these
ascii representations of coding language. Much of this bickering "lets
use this character for that", etc. seems futile, while we all have a
similar idea of the code tree to generate. Infact modern compilers
work that way, first parse the language in a tree, then create
assembler code. So why no environment that lets you edit the tree
directly and make it look like the way you want? I coded something
with GUI using Remote Procedure CALLs to interact with the tree, but
gave up when working with it all was unbareable slow. Other problems
would be incompatibility with versioning system, or the unability to
provide snippets in emails. Today i think it aint so bad we use ascii
files as a common exchange format. And with dynamic language and
compiling generated code while it runs, it gets even further off. The
only thing that left is my strict use of tabs as indentation, and only
as indentation so the viewer can adjust its width to personal liking.

The other view, I'm no fan of static inspecting. Hasn't this be one of
the major aspect in theoreticals informatics, to be able to proof a
piece of code is correct? It led us nowhere, or did it? Despite this,
these GUIs giving you hints and calling for common errors are sure
important for the ecosystem. The ecosystem is what in my opinion is
more important for a lanauge anyway than its implementation detail X
or Y. I want to be able to just copy/paste any error from the
commandline and find docends of google hits of people using this
language and telling me what I did wrong.