Re: behaviour of assert on non-string errors

[Andrew Starks <andrew.starks@...>]

On Saturday, January 18, 2014, Dirk Laurie <dirk.laurie@gmail.com> wrote:
> 2014/1/18 Philipp Janda <siffiejoe@gmx.net>:
>
> > I think we already had this discussion a year ago[1].
> >   [1]: http://lua-users.org/lists/lua-l/2013-01/msg00084.html
>
> So we did, and certain people still think the same way as they
> did then. :-)

A discussion on what is best might make for some good sport. And also it is true: people often use assert as short hand for:

if not x then error("you must supply x." ) end 

...which, in fact, is a paraphrase of the example in PiL and that is as close as it gets to "blessing" an idiom as the is in Lua. 

So then there is the question of changing a behavior that others rely on. Don't see that here... Adding complexity to the language? Nope. Making things harder for those that don't see a need for this flexibility? I don't think so. And on and on. 

But then there is the question of complexity in Lua's implementation, and the added tests and maintenance required. 

So, instead of saying that it should be in there, my opinion is that if if were, it'd be valuable to me and the slow-down that would happen upon erroring would not at all be something that would matter. That said, I often wrap assert and don't really use it in the repl.

-Andrew