Re: Dubiousness of the debug library

[Daurnimator <quae@...>]

On 9 July 2015 at 15:40, Dirk Laurie <dirk.laurie@gmail.com> wrote:
> 2015-07-09 5:57 GMT+02:00 Soni L. <fakedme@gmail.com>:
>> On 08/07/15 11:12 PM, Luiz Henrique de Figueiredo wrote:
>>>
>>> Use
>>>         debug.setmetatable(nil, {__index = {}})
>>>
>> I should make a library for this that doesn't require debug.
>
> Why are many Lua programmers reluctant to use the debug library?
>
> Lua 5.3.1 Reference Manual on the debug library:
>
> | This library provides the functionality of the debug interface (§4.9)
> | to Lua programs. You should exert care when using this library.
> | Several of its functions violate basic assumptions about Lua code
> | (e.g., that variables local to a function cannot be accessed from
> | outside; that userdata metatables cannot be changed by Lua code;
> | that Lua programs do not crash) and therefore can compromise
> | otherwise secure code. Moreover, some functions in this library
> | may be slow.
>
> This is daunting, certainly, but it used to be worse.
>
> Lua 5.1.5 Reference Manual on the debug library:
>
> | The functions provided here should be used exclusively for debugging
> | and similar tasks, such as profiling. Please resist the temptation to use
> | them as a usual programming tool
>
> By reading these together, one can draw the conclusion that the Lua
> developers no longer feel that the debugging library is only there for
> debugging, profiling etc, and that it is now OK to use its functions
> as
> a usual programming tool provided that you exert care.
>
> So why are we still squeamish about it?
>

It's sort of a 'here be dragons' indication:
Once the debug library is in use, it can break many assumptions.

Furthermore, in this *particular* instance, a global behaviour change
is being made.
Any well made library should generally have no global side effects.