Re: Arguments of dofile

[Roberto Ierusalimschy <roberto@...>]

> > Could some of the old-timers on this list recap its effectiveness
> > as an influence on language design?  Which ideas currently in Lua 
> > were first aired here?
> 
> See the HOPL paper. It mentions the adoption of the MIT license (not a
> language design issue but an important issue nevertheless), coroutines,
> and lexical scoping, for instance.

Just to name a few more, after Lua 5.2 alpha we got ellimination
of loadin (merged into load), tonumber working with large numbers,
searchpath preserving dots, and applying 'tostring' to arguments of %s
(still being implemented).

Actually, a lot of ideas currently in Lua were first aired in this list.
Many others were the product of related discussions in this list. But we
judge them by quality, not by popularity.

We wellcome suggestions, but I must say that this list recently has had
too many of them, many of them of low quality. (I am not talking about
this one.) People suggesting something should do their homework. If they
expect us to spend some time considering the idea, they should spend
more time considering it. (That includes learning Lua properly, cheking
for alternatives, etc.)

Among other things, a good suggestion should consider:

- What problem it is trying to solve.
- Whether the proposal actually solves the problem.
- Whether it solves problems other than your particular one.
- Whether there is already a good way to do what it achieves.
- Whether it is really implementable. (Unless the implementation is
trivial, you should implement it before doing a suggestion. When in
doubt, it is not trivial. Check performance with care.)
- Whether it is simple to explain/document. (The real problem with
current '#'?)

-- Roberto