RFC: Deprecating the Module Operator

[Frits van Bommel]

%u wrote:
> The current module operator, the ".", seems to be a bad choice,

I presume you mean the prefix-'.', not the binary-'.' "member-of" operator.

> because it is easily overseen and in conjunction with an accidentically
> deleted character may result in bugs, that are hard to discover:
>   var= o1*.o2;
> vs.
>   var= o1..o2;
> vs.
>   var=.o1.o2;

When is the first variant valid syntax?
I'm pretty sure the second one will also never be valid syntax. At 
least, not for D 1.0.
The only potential conflict like 2 vs 3 that I can see will be inside 
array brackets[1], i.e. "arr[o1..o2]" vs "arr[.o1.o2]". And I expect 
that will be easily distinguished by the resulting compile error on the 
one that doesn't make sense for the type of symbol o1 is...


[1]: Not necessarily on an array, due to operator overloading, but you 
know what I mean.


> Moreover the module operator just seems to be syntax sugar, because
> one needs not to write the name of the module.

You do need to if there's otherwise a name conflict. (Though that can 
also be resolved by alias usage, it's sometimes cleaner to just use the 
full name)

> OTOH if there is a need not to write the module name, then the well
> known and more general file hierarchy navigation notation should be
> used, where "//" denotes the root and "../" denotes direct ancestors.

Please not.
I have no words for this. So I'll take this quote about D from "Jeff M": 
"It's like C++ except that it doesn't make me want to shove knitting 
needles into my eyes." and just add this: let's keep it that way :P.

Luckily '//' is used for line comments in D so it's not going to happen.

> --
> Anonymity is not a Crime

In a public forum though, pseudonymity is generally preferred. That way 
you can at least tell people apart ;).