www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - Is !in planed?

reply Dawid =?ISO-8859-2?Q?Ci=EA=BFarkiewicz?= <arael fov.pl> writes:
I can't find a thread with the decision about !in.
-- 
Dawid Ciężarkiewicz | arael
Jun 25 2005
parent reply Sean Kelly <sean f4.ca> writes:
In article <d9kq6l$17gq$1 digitaldaemon.com>, Dawid
=?ISO-8859-2?Q?Ci=EA=BFarkiewicz?= says...

I can't find a thread with the decision about !in.


Now that we have !is I don't see why not.  Though the usage would be a bit odd
consdiering 'in' returns a pointer.


Sean
Jun 25 2005
parent reply Dawid =?ISO-8859-2?Q?Ci=EA=BFarkiewicz?= <arael fov.pl> writes:
Sean Kelly wrote:

 Now that we have !is I don't see why not.  Though the usage would be a bit
 odd consdiering 'in' returns a pointer.


Yes. Despite the fact IMO the compiler should take care about double lookup
when using 'in' - not the programer - '!in" keyword has nothing to return
but bool so should be very natural addition do language (it will be more
natural than custom 'in' ;-) ).

So, should I think:
assert("!in" in D);
?

-- 
Dawid Ciężarkiewicz | arael
Jun 26 2005
parent reply Sean Kelly <sean f4.ca> writes:
In article <d9m711$2mcb$1 digitaldaemon.com>, Dawid
=?ISO-8859-2?Q?Ci=EA=BFarkiewicz?= says...

Sean Kelly wrote:

 Now that we have !is I don't see why not.  Though the usage would be a bit
 odd consdiering 'in' returns a pointer.


Yes. Despite the fact IMO the compiler should take care about double lookup
when using 'in' - not the programer - '!in" keyword has nothing to return
but bool so should be very natural addition do language (it will be more
natural than custom 'in' ;-) ).

So, should I think:
assert("!in" in D);


What I'd like is this:

int[int] aa;
int* val;
if( 5 in aa ) // 'in' returns bit
if( 5 !in aa ) // '!in' returns bit
if( ( val = aa[5] ) != null ) // rvalue subscript returns ptr but no insert
val = ( aa[5] = 6 ); // lvalue subscript does insert and returns ptr


Sean
Jun 26 2005
next sibling parent reply Chris Sauls <ibisbasenji gmail.com> writes:
Sean Kelly wrote:

 What I'd like is this:
 
 int[int] aa;
 int* val;
 if( 5 in aa ) // 'in' returns bit


This works currently, as a null pointer is a 'false' value in D.


 if( 5 !in aa ) // '!in' returns bit
 if( ( val = aa[5] ) != null ) // rvalue subscript returns ptr but no insert


This might be nifty-ish.  Currently it throws an exception
(ArrayBoundsException) so you 
could "just" mimic this if/if-else with a try block.








I never said it was pretty...


 val = ( aa[5] = 6 ); // lvalue subscript does insert and returns ptr


I dunno.  Its a consistancy thing, really I guess.  In almost all other places,
the value 
of an assignment is not a pointer.  (Pointer typed variables not withstanding.)
 I don't 
think I would ever remember to dereferance AA assignments.  Well no, I guess
I'd learn 
after a few weeks of fighting the compiler about it... but still.

-- Chris Sauls
Jun 26 2005
parent Sean Kelly <sean f4.ca> writes:
In article <d9mu5u$40c$1 digitaldaemon.com>, Chris Sauls says...

 val = ( aa[5] = 6 ); // lvalue subscript does insert and returns ptr


I dunno.  Its a consistancy thing, really I guess.  In almost all other places,
the value 
of an assignment is not a pointer.  (Pointer typed variables not withstanding.)
 I don't 
think I would ever remember to dereferance AA assignments.  Well no, I guess
I'd learn 
after a few weeks of fighting the compiler about it... but still.


Unless we get reference types I think returning a pointer makes the best sense
:p  Though I agree it's a tad odd.


Sean
Jun 26 2005
prev sibling parent reply "Regan Heath" <regan netwin.co.nz> writes:
On Sun, 26 Jun 2005 16:22:17 +0000 (UTC), Sean Kelly <sean f4.ca> wrote:

 In article <d9m711$2mcb$1 digitaldaemon.com>, Dawid
 =?ISO-8859-2?Q?Ci=EA=BFarkiewicz?= says...

 Sean Kelly wrote:

 Now that we have !is I don't see why not.  Though the usage would be a  
 bit
 odd consdiering 'in' returns a pointer.


 Yes. Despite the fact IMO the compiler should take care about double  
 lookup
 when using 'in' - not the programer - '!in" keyword has nothing to  
 return
 but bool so should be very natural addition do language (it will be more
 natural than custom 'in' ;-) ).

 So, should I think:
 assert("!in" in D);


 What I'd like is this:

 int[int] aa;
 int* val;
 if( 5 in aa ) // 'in' returns bit
 if( 5 !in aa ) // '!in' returns bit
 if( ( val = aa[5] ) != null ) // rvalue subscript returns ptr but no  
 insert
 val = ( aa[5] = 6 ); // lvalue subscript does insert and returns ptr


What I'd like is:

int[int] aa;
int val;

if (5 in aa)  //'in' returns bit
if (5 !in aa) //'!in' returns bit
if (aa.contains(5,val)) //contains returns bit, assigns val if present
aa[5] = val = 6;

It would also make sense to have the alternate form of contains:
   if (aa.contains(5))     //contains returns bit

even if it's identical to 'in' in nature. Or perhaps we keep 'in' as it  
currently is, returning a pointer. I know I won't be using it (desire to  
avoid pointers whereever possible).

If however 'in' remains as it currently is, I can write myself the 2  
contains routines. I can even write them as templates, however without  
implicit template instantiation the calls look like:

if (contains!(int)(aa,5))
if (contains!(int)(aa,5,val))

It would be nice if the form matched the calls above. I can't recall.. it  
might already be possible to use:

if (aa.contains!(int)(5))
if (aa.contains!(int)(5,val))

but it's not quite perfect even so.

Regan
Jun 26 2005
parent reply Chris Sauls <ibisbasenji gmail.com> writes:
Regan Heath wrote:

 if (aa.contains!(int)(5))
 if (aa.contains!(int)(5,val))


I don't think this works... yet, anyhow.  Although you could do:




-- Chris Sauls
Jun 26 2005
parent "Regan Heath" <regan netwin.co.nz> writes:
On Sun, 26 Jun 2005 17:45:50 -0500, Chris Sauls <ibisbasenji gmail.com>  
wrote:

 Regan Heath wrote:

 if (aa.contains!(int)(5))
 if (aa.contains!(int)(5,val))


 I don't think this works... yet, anyhow.  Although you could do:





Really? I can't seem to make that happen in a generic way. :(
I assume you'd use 'alias' in the template.

Regan
Jun 26 2005