Import improvement

[Chris Nicholson-Sauls]

On Monday, 15 October 2012 at 15:37:06 UTC, Andrei Alexandrescu 
wrote:
> I don't think imports from a specific package have been 
> considered.
>
> In my personal opinion, imports are a necessary evil and it's 
> sort of a bummer that the most accessible place in any source 
> file - the top lines - is occupied by the crappy legal 
> disclaimer (which, after having talked to a lawyer, I always 
> put at the bottom since being at the top is not a requirement), 
> and the litany of imports that the module is using. I'd make 
> all imports local or put them at the bottom of the file if it 
> weren't too much of a shock to others.
>
> Three remarks on this particular problem.
>
> 1. I expect large packages to introduce a module "all.di" or 
> "_.di" to publicly import everything in the package. That could 
> help some use cases.

It is a common practice (usually all.di) but perhaps it could 
help to establish an official convention.  Nothing in the 
language, just the styleguide.  (I know this has already come up 
and been discussed.)

> 2. The import declaration accepts a list of modules, and 
> several may be on one line. I think that's a significant 
> positive difference from C, C++, or Go, all of which force one 
> imported module per line. I always advocate imports from the 
> same package in the same "import" declaration, ordered 
> alphabetically:
>
> import fuji.filesystem, fuji.font, fuji.material, fuji.matrix,
>     fuji.primitive, fuji.render, fuji.system;
> import std.algorithm, std.conv, std.random, std.string, std.xml;
> ...
>
> That makes the existing system much more palatable.

I've done this very thing for eons, and yes you are quite right!  
(Although my formatting is different, but I digress.)  That said, 
I think the OP still has a valid, even quite strong point.  
Consider, in comparison to your sample:

import fuji.( filesystem, font, material, matrix, primitive, 
render, system );
import std.( algorithm, conv, random, string, xml );

Certainly less visual noise.  An even *better* level of 
improvement comes if we provide an alternate and similar syntax 
for selective imports, so that they no longer have to be 
separated.

import std.( algorithm, conv, random, xml,
     stdio:( write, writef, writefln ),
     string:( munch, splitLines, xformat, xsformat )
);

As for implementation, how difficult is it for the compiler to 
internally expand something like this into the traditional litany?

> 3. I think local imports are currently underutilized. It would 
> be interesting to see what kind of project dynamics they enable.

I can agree with this.

-- Chris Nicholson-Sauls