Re: Is string always same as number? (the consistency of lua)

[steve donovan <steve.j.donovan@...>]

On Thu, Feb 10, 2011 at 12:30 PM, Axel Kittenberger <axkibe@gmail.com> wrote:
> cool sounding title would be "touch typing, thought typing, tough
> typing! An Analysis of t-typing types"

Good background reading:

http://blog.steveklabnik.com/what-to-know-before-debating-type-systems-0

Nice overview of some of the myths (e.g. that statically typed
programs need more actual typing - but implicit static typing takes
away a lot of the bookkeeping pain)

He mentions that many like the REPL available in dynamic languages,
but indeed, that's an engineering question. Pretty straightforward to
do a C# REPL with the runtime compiler facilities, and I myself did a
C++ interpreter in my youth.

There's the position advocated by Martin Fowler, which is (a) strong
typing has real costs (long build times ,etc) and so (b) rather go for
strong testing than strong typing.

The relevance to Lua: it's fine as it is. Paranoia has its place in
programming and one can be paranoid in Lua, but the extra type checks
are expensive at _runtime_ instead.

Another point of relevance: it is often said that static typing is a
necessary means to get serious performance. Well, we need only point
to LuaJIT to show that this is also a myth.

steve d.