Re: A guide to building Lua modules

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

On Wed, Apr 16, 2014 at 11:05 PM, Petite Abeille
<petite.abeille@gmail.com> wrote:
> "About the list behaviour”
> — André Carregal, 2005
>
> http://article.gmane.org/gmane.comp.lang.lua.general/13580

Nice one, PA!  Should be required introductory reading. It's a good
engineering analysis of group behaviour - whether a desired feature
raises enough heat for the tweakers to start proof-of-concept.  And
describes the "long tail" of irritated messages when it comes clear
that someone's cherished idea has no chance in hell.

I particularly like the typo "mudules" which means "a module with no
clearly defined central purpose" -  rule #1! [1]

As for "anti-featurism", well it may come across like that, but people
here are at least polite about it.  All features have a cost, whether
conceptually or in performance.[2]

The only feature proposal (recycled) in the last two weeks that I
could get behind is "local x,y,z in t" - well-defined sugar for a
common irritating case.

[1] although, Penlight is Not a Module. But the idea of a
one-stop-shop for a collection of Useful Utilities is not as
attractive to the community as I imagined.  For one thing, it makes
discoverability harder; if PL were a dozen separate rocks they would
all have their own description in the manifest. I think the Readme
should become a catalogue, not a rationale...

[2] a Lua fork in the hands of hotheads would take about a year to
become a blurred, slow mess. Better to have forks with a Benevolent
Dictator (or Triumvirate).  I'd consider Moonscript to be a successful
example of that, since it is consistently designed (although man those
\ method calls are ugly...)