Decrease chance of error and improve maintenance

[paulo-ferraz-oliveira]

I found different defaults in elvis_style and elvis_rulesets, and in those cases the public Wiki won.

[Fix inaka/elvis#525].

[paulo-ferraz-oliveira]

This isn't yet complete, but if the general direction isn't this one, I'll want to adapt before I move to the other rulesets.

[paulo-ferraz-oliveira]

Anyway, I had a local version that used macros for functions default/2 and option. It simplified stuff, but may not be what you wanted (I also find it borderline convoluted).

Basically, where you have

option(OptionName, RuleConfig, Rule)

you'd have

?OPT(OptionName)

with

-define(OTP(OptionName), (fun () -> maps:get(OptionName, RuleConfig, default(?FUNCTION_NAME, OptionName)) end))

(with the assumption that RuleConfig was defined in the call context - I know, "dirty").

[paulo-ferraz-oliveira]

I then added a further change where default(?FUNCTION_NAME, OptionName) would also become

-define(?DEF(OptionName), OptionName => default(?FUNCTION_NAME, OptionName)).
% so we could build the options maps as:
#{ ?DEF(limit) },

[paulo-ferraz-oliveira]

Why these (local) changes?

I just feel that defining stuff as

    #{ limit => default(god_modules, limit)
     , ignore => default(god_modules, ignore)

is more error-prone than (though the advantage is there's no "magic")

    #{ ?DEF(limit)
     , ?DEF(ignore) }

and also I'd prefer to reduce stuff like

    Regex = option(regex, RuleConfig, function_naming_convention),
    Ignores = option(ignore, RuleConfig, function_naming_convention),

to

    Regex = ?OPT(regex),
    Ignores = ?OPT(ignore),

thus removing the need to keep using RuleConfig and the function name (or ?FUNCTION_NAME).

[paulo-ferraz-oliveira]

I pushed a preparation commit for those (macro-based) changes should we want them in elvis_style.

[elbrujohalcon]

Oh, and please move us to meck 0.9.0 🙏

[elbrujohalcon]

Sorry, @paulo-ferraz-oliveira … but we have a rule against macros.
I'm inclined to reluctantly allow ?FUNCTION_NAME… although I would not use that myself. But that's like… the farthest I'm willing to go in that direction.

[paulo-ferraz-oliveira]

Oh, and please move us to meck 0.9.0 🙏

I didn't want to change it, since it wasn't affecting the scope of my PR. But I'll do it, as per your request.

Edit: 317c0d5

[paulo-ferraz-oliveira]

Sorry, @paulo-ferraz-oliveira … but we have a rule against macros.
I'm inclined to reluctantly allow ?FUNCTION_NAME… although I would not use that myself. But that's like… the farthest I'm willing to go in that direction.

Hm... I guess it never made it to elvis. It is proposed here though. We'll eventually get there :)

Edit: also wanted to point out that elvis already has quite a few user-defined macros; and not that's it's "better", but ?FUNCTION_NAME is built-in (since Erlang/OTP 19, if I'm not mistaken).

Edit: 59300dd

[paulo-ferraz-oliveira]

@elbrujohalcon, don't merge just yet; there's stuff missing for elvis_project. (I just wanted to understand if I was going in the right direction)

Edit: a7f57b8

[elbrujohalcon]

Is it ready now, @paulo-ferraz-oliveira ?

[paulo-ferraz-oliveira]

It is.