• Era Scarecrow (56/56) Oct 16 2012 I haven't found this specific topic anywhere in the archives so
• ixid (2/7) Oct 16 2012 This seems sensible. Multiple with seems like a recipe for
• Era Scarecrow (37/46) Oct 17 2012 True... But if you're planning on going multiple levels of with,

"Era Scarecrow" <rtcvb32 yahoo.com> writes:

  I haven't found this specific topic anywhere in the archives so 
I'll throw it out there for feedback.

[quote] TDPL pg. 81
  "There is no ambiguity-related danger in using nested 'with's 
because the language disallows shadowing of a symbol introduced 
by an outer with by a symbol introduced by an inner with. In 
brief, in D a local symbol can never shadow another local symbol"
[/quote]

  Technically 'with' blocks can be placed just about anywhere you 
could run normal code (and sometimes replaces blocks), however 
there are cases where I'm required to use extra braces. Mind you 
this is minor syntactical issues, but..

   enum E {}
   enum F {}
   enum G {}

currently:

   void func()
   //in/out contracts
   body {
     with (E) {

     }
   }

block replacement:

   if () {
   } else with(E) {
     //
   }

Theoretically legal...

   void func()
   //in/out contracts
   body with (E) { //with replaces normal block

   }

  The above refuses to compile, however I don't feel I'd need an 
extra level of indentation, and since it's static data like Enums 
perhaps you'd want multiples. Last 'with' only accepts one 
argument, but I wonder if it wouldn't hurt to enter multiple. 
Mind you it will still error during compiling if there's 
ambiguity. This makes more sense with enums and statically known 
data vs variables.

   with(E) {
     with(F) {
       with(G) {
         //code
       }
     }
   }

or (better, TDPL pg. 81)

   with(E) with(F) with(G) {
     //code
   }

vs

   with(E, F, G) {
     //code
   }

  Perhaps a feature request. I know it's not essential so I won't 
try and push it, but syntactical sugar can't hurt right?

"ixid" <nuaccount gmail.com> writes:

Theoretically legal...

  void func()
  //in/out contracts
  body with (E) { //with replaces normal block

  }

This seems sensible. Multiple with seems like a recipe for 
confusion and member name clashes.

"Era Scarecrow" <rtcvb32 yahoo.com> writes:

On Wednesday, 17 October 2012 at 06:07:34 UTC, ixid wrote:
Theoretically legal...

 void func()
 //in/out contracts
 body with (E) { //with replaces normal block

 }

 This seems sensible. Multiple with seems like a recipe for 
 confusion and member name clashes.

  True... But if you're planning on going multiple levels of with, 
then it isn't going to matter much.

  In a project of mine I'll have a few different enum types, each 
one with their own sets of flags used in a structure. So you're 
looking at something like.. As an off the wall example.

   enum OnOff {off, on}
   enum Speed {stop, slow, medium, fast, reallyFast, lightspeed}
   enum MemoryType {allocated, stack}
   enum EngineType {internalCombustion, electrical, magical, warp, 
teleportion}
   enum Size {tiny, small, medium, large, huge}

   struct Vehicle {
     OnOff state;
     Speed speed;
     const MemoryType memAt;
     const EngineType engine;
     const Size size;
   }

  So in this kind of case where none of the enums clash, then 
manually doing them. True there's a clash from size/speed, but 
that's fairly easy to fix in the case it's used.

   with (OnOff, Speed, MemoryType, EngineType, Size)
//    or
//  with (OnOff) with(Speed) with(MemoryType)
//      with(EngineType) with(Size)
   {
     Vehicle Enterprise = {off, stop, stack, warp, huge};

     //fully explicit, but doesn't buy you much except clutter
     Vehicle Enterprise2 = {OnOff.off, Speed.stop, 
MemoryType.stack,
                             EngineType.warp, Size.huge};

     //wrong order, enums complain but is fairly obvious
     //based on types (and the error message) how to reorder it.
     Vehicle ModelT = {stop, allocated, medium, 
internalCombustion, off};
   }