www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Adding finally to switch

reply Jesse Phillips <jessekphillips gmail.com> writes:
I haven't given it much thought, but I figured I'd let some other people 
look at it too.

Switch statements are nice, many people hate having to use break; all the 
time, but I don't and am not interest in the debate. What I think is 
missing from a switch statement is a finally section. Most of the time I 
don't have a use for the fall-through feature of switch, but I do have a 
use for doing one or more things that are the same in every case.

As I haven't given it a lot of thought I will leave out some constraint 
ideas, and just see other peoples thoughts. I don't think it would ruin 
compatibility of any sort (backwards or C).
Mar 31 2008
next sibling parent BCS <ao pathlink.com> writes:
Reply to Jesse,


 I haven't given it much thought, but I figured I'd let some other
 people look at it too.
 
 Switch statements are nice, many people hate having to use break; all
 the time, but I don't and am not interest in the debate. What I think
 is missing from a switch statement is a finally section. Most of the
 time I don't have a use for the fall-through feature of switch, but I
 do have a use for doing one or more things that are the same in every
 case.
 
 As I haven't given it a lot of thought I will leave out some
 constraint ideas, and just see other peoples thoughts. I don't think
 it would ruin compatibility of any sort (backwards or C).
 


this could be made to work:

|import std.stdio;
|
|void main()
|{
|  switch(1)
|  {
|    scope(success) writef("Finaly\n");
|
|    case 1: writef("first\n"); break;
|  }
|}

it doesn't work because it's implemented the same way exceptions are (I think) 
but doesn't need to be.
Mar 31 2008
prev sibling next sibling parent reply "Scott S. McCoy" <tag cpan.org> writes:
My thought on switch has always been:  If you're using it as multiple if
statements, you're misusing it.

I always looked at switch as an opportunity to optimize, little more.  I
never really thought the idea of switching on strings or similar was
really that sweet.  If you ask me, such a construct is a fine idea, but
we should create a different construct for it (and a dispatch table
typically works better for that type of thing).  That being said, switch
is not a tool you use every day.  But that's fine, because when you
really *need* a real switch, its so incredibly handy it's unreal, and
you become grateful it's there and that it's so fast for what it does,
and that it lets you gracefully fall through and create "groups" of
behavior where you can't do that with other constructs as easily.

Cheers,
    Scott S. McCoy

On Tue, 2008-04-01 at 03:56 +0000, Jesse Phillips wrote:

 I haven't given it much thought, but I figured I'd let some other people 
 look at it too.
 
 Switch statements are nice, many people hate having to use break; all the 
 time, but I don't and am not interest in the debate. What I think is 
 missing from a switch statement is a finally section. Most of the time I 
 don't have a use for the fall-through feature of switch, but I do have a 
 use for doing one or more things that are the same in every case.
 
 As I haven't given it a lot of thought I will leave out some constraint 
 ideas, and just see other peoples thoughts. I don't think it would ruin 
 compatibility of any sort (backwards or C).
Mar 31 2008
parent reply Jesse Phillips <jessekphillips gmail.com> writes:
On Mon, 31 Mar 2008 21:52:17 -0700, Scott S. McCoy wrote:


 My thought on switch has always been:  If you're using it as multiple if
 statements, you're misusing it.


I would agree, except I don't. :) I thought one of the goals in 
programming was to choose tools that would reduce the number of repeats 
in code. That is to say if I have several things X could be and want to 
do something different for all of them, would it be logical to only have 
to write X once instead of if(X == B) else if (X == C) ...?

Anyway maybe BCS has a good choice for it.
Apr 01 2008
parent Yigal Chripun <yigal100 gmail.com> writes:
Jesse Phillips wrote:

 On Mon, 31 Mar 2008 21:52:17 -0700, Scott S. McCoy wrote:

   

 My thought on switch has always been:  If you're using it as multiple if
 statements, you're misusing it.
     


 I would agree, except I don't. :) I thought one of the goals in 
 programming was to choose tools that would reduce the number of repeats 
 in code. That is to say if I have several things X could be and want to 
 do something different for all of them, would it be logical to only have 
 to write X once instead of if(X == B) else if (X == C) ...?

 Anyway maybe BCS has a good choice for it.
   


except, you're forgetting one important issue - multi threading:
if you're on a single processor/core/PC/etc, switch would be faster then
a series of if statements, as it could be implemented as a table.
however, on a multi-core/mutli-cpu the exact opposite is true, and it's
more general as in the next snippet:
if (x == 1) do_a();
if (x > 0) do_b();
if (x < 2) do_c();
the above could be run in parallel (providing of course, that do_a, do_b
and do_c are independent) on different cores and it doesn't have to be
deterministic (only one case is run).
in new code meant to be run on multi-core PCs the above is much better
than a switch.

--Yigal

PS - writing the above made me thinking: do we need a special
parallel_switch syntax for code such as the above where all cases are
independent from each other and can run in parallel?
Apr 01 2008
prev sibling next sibling parent reply downs <default_357-line yahoo.de> writes:
Jesse Phillips wrote:

 I haven't given it much thought, but I figured I'd let some other people 
 look at it too.
 
 Switch statements are nice, many people hate having to use break; all the 
 time, but I don't and am not interest in the debate. What I think is 
 missing from a switch statement is a finally section. Most of the time I 
 don't have a use for the fall-through feature of switch, but I do have a 
 use for doing one or more things that are the same in every case.
 
 As I haven't given it a lot of thought I will leave out some constraint 
 ideas, and just see other peoples thoughts. I don't think it would ruin 
 compatibility of any sort (backwards or C).


This can be generalized.

I would ask to introduce the keyword "then" to indicate "this is run after the
previous construct has successfully completed, executing at least one branch."

It could be applicable to [do/]while-loops as well as switches.

Finally is an already existing keyword, which could lead to problems with
parsing.

 --downs
Mar 31 2008
parent downs <default_357-line yahoo.de> writes:
downs wrote:

 Jesse Phillips wrote:

 I haven't given it much thought, but I figured I'd let some other people 
 look at it too.

 Switch statements are nice, many people hate having to use break; all the 
 time, but I don't and am not interest in the debate. What I think is 
 missing from a switch statement is a finally section. Most of the time I 
 don't have a use for the fall-through feature of switch, but I do have a 
 use for doing one or more things that are the same in every case.

 As I haven't given it a lot of thought I will leave out some constraint 
 ideas, and just see other peoples thoughts. I don't think it would ruin 
 compatibility of any sort (backwards or C).


 
 This can be generalized.
 
 I would ask to introduce the keyword "then" to indicate "this is run after the
previous construct has successfully completed, executing at least one branch."
 
 It could be applicable to [do/]while-loops as well as switches.
 
 Finally is an already existing keyword, which could lead to problems with
parsing.
 
  --downs


I am stupid. If the OP was an april fools, I fell for it hook, line and sinker.

I had completely forgotten that switch already covers all cases.

Ignore this keyword for switches please, folks.

(I still think it's a good idea for loops)

 --downs
Mar 31 2008
prev sibling parent reply Henning Hasemann <hhasemann web.de> writes:
Jesse Phillips <jessekphillips gmail.com> wrote:

 I haven't given it much thought, but I figured I'd let some other
 people look at it too.
 
 Switch statements are nice, many people hate having to use break; all
 the time, but I don't and am not interest in the debate. What I think
 is missing from a switch statement is a finally section. Most of the
 time I don't have a use for the fall-through feature of switch, but I
 do have a use for doing one or more things that are the same in every
 case.


I don't think I get your point. If you have code that is the same in
*every* case why cant you do:

do_this_before_each_case();
switch(foo) {
 // ...
}
do_this_after_each_case();

Henning

-- 
GPG Public Key:
http://gpg-keyserver.de/pks/lookup?op=get&search=0xDDD6D36D41911851
Apr 01 2008
next sibling parent reply Bill Baxter <dnewsgroup billbaxter.com> writes:
Henning Hasemann wrote:

 Jesse Phillips <jessekphillips gmail.com> wrote:

 I haven't given it much thought, but I figured I'd let some other
 people look at it too.

 Switch statements are nice, many people hate having to use break; all
 the time, but I don't and am not interest in the debate. What I think
 is missing from a switch statement is a finally section. Most of the
 time I don't have a use for the fall-through feature of switch, but I
 do have a use for doing one or more things that are the same in every
 case.


 
 I don't think I get your point. If you have code that is the same in
 *every* case why cant you do:
 
 do_this_before_each_case();
 switch(foo) {
  // ...
 }
 do_this_after_each_case();
 
 Henning
 


Maybe he meant every case except for the default?  Still not that big a 
deal, though.  Something like this works:

bool passed=false;
switch(foo) {
    ...
   default:
      passed = true;
}
do_this_after_each_non_default_case();

--bb
Apr 01 2008
parent Bill Baxter <dnewsgroup billbaxter.com> writes:
Bill Baxter wrote:

 Henning Hasemann wrote:

 Jesse Phillips <jessekphillips gmail.com> wrote:

 I haven't given it much thought, but I figured I'd let some other
 people look at it too.

 Switch statements are nice, many people hate having to use break; all
 the time, but I don't and am not interest in the debate. What I think
 is missing from a switch statement is a finally section. Most of the
 time I don't have a use for the fall-through feature of switch, but I
 do have a use for doing one or more things that are the same in every
 case.


 I don't think I get your point. If you have code that is the same in
 *every* case why cant you do:

 do_this_before_each_case();
 switch(foo) {
  // ...
 }
 do_this_after_each_case();

 Henning


 
 Maybe he meant every case except for the default?  Still not that big a 
 deal, though.  Something like this works:
 
 bool passed=false;
 switch(foo) {
    ...
   default:
      passed = true;
 }


// Doh! Works better if you remember to check the condition
if(!passed)

 do_this_after_each_non_default_case();



-bb
Apr 01 2008
prev sibling parent Jesse Phillips <jessekphillips gmail.com> writes:
On Tue, 01 Apr 2008 10:20:04 +0200, Henning Hasemann wrote:


 Jesse Phillips <jessekphillips gmail.com> wrote:

 I haven't given it much thought, but I figured I'd let some other
 people look at it too.
 
 Switch statements are nice, many people hate having to use break; all
 the time, but I don't and am not interest in the debate. What I think
 is missing from a switch statement is a finally section. Most of the
 time I don't have a use for the fall-through feature of switch, but I
 do have a use for doing one or more things that are the same in every
 case.


 
 I don't think I get your point. If you have code that is the same in
 *every* case why cant you do:
 
 do_this_before_each_case();
 switch(foo) {
  // ...
 }
 do_this_after_each_case();
 
 Henning


Well, I was thinking there might be special constraints added to it, like 
Bill's example, where default is left out. But as he should I guess in 
isn't too hard to make your own. To which I say never mind.
Apr 01 2008