Hi there,

I have a suggestion for an addition to the foreach syntax. Even though D might not get it implemented, I'm very interested in your commments. So here it is: How about a "no-items" clause?

I've thought about which keyword to use, and came up with "else" (which is already a keyword), "ornone," and "otherwise," which are unlikely to be identifiers anyway. Allow me to demonstrate:

// Currently:
string[] matches = r.match(expr);

if (matches.length == 0) {
// There are no items.
// Do some stuff.
} else {
foreach (string m; matches) {
// There are items.
// Do some other stuff.
}
}

// With suggestion:
foreach (string m; r.match(expr)) {
// There are items.
// Do some stuff.
} else {
// There are no items.
// Do some other stuff.
}

What do you guys think?
--AJG.


==========================
if (!sin) throw (rock[0]);

I never would've thought of it... but I do kinda like it.  The (if !length else foreach) pattern does happen a good deal, and this does shrink it down quite a lot, without any fuss.  Probably fairly straightforward to implement, as well.  The only catch I see right off, is how this would operate in terms of opApply?  Maybe a special return value set aside to mean, "I am empty so run the else clause."

-- Chris Sauls

AJG wrote:
> Hi there,
> 
> I have a suggestion for an addition to the foreach syntax. Even though D might
> not get it implemented, I'm very interested in your commments. So here it is:
> How about a "no-items" clause?
> 
> I've thought about which keyword to use, and came up with "else" (which is
> already a keyword), "ornone," and "otherwise," which are unlikely to be
> identifiers anyway. Allow me to demonstrate:
> 
> // Currently:
> string[] matches = r.match(expr);
> 
> if (matches.length == 0) {
> // There are no items.
> // Do some stuff.
> } else {
> foreach (string m; matches) {
> // There are items.
> // Do some other stuff.
> }
> }
> 
> // With suggestion:
> foreach (string m; r.match(expr)) {
> // There are items.
> // Do some stuff.
> } else {
> // There are no items.
> // Do some other stuff.
> }
> 
> What do you guys think?
> --AJG.
> 
> 
> ==========================
> if (!sin) throw (rock[0]);

In article <d8to07$ikt$1@digitaldaemon.com>, Chris Sauls says...
>
>I never would've thought of it... but I do kinda like it.  The (if !length else foreach) pattern does happen a good deal, and this does shrink it down quite a lot, without any fuss.  Probably fairly straightforward to implement, as well.  The only catch I see right off, is how this would operate in terms of opApply?  Maybe a special return value set aside to mean, "I am empty so run the else clause."
>
>-- Chris Sauls
>
>AJG wrote:
>> Hi there,
>> 
>> I have a suggestion for an addition to the foreach syntax. Even though D might not get it implemented, I'm very interested in your commments. So here it is: How about a "no-items" clause?
>> 
>> I've thought about which keyword to use, and came up with "else" (which is already a keyword), "ornone," and "otherwise," which are unlikely to be identifiers anyway. Allow me to demonstrate:
>> 
>> // Currently:
>> string[] matches = r.match(expr);
>> 
>> if (matches.length == 0) {
>> // There are no items.
>> // Do some stuff.
>> } else {
>> foreach (string m; matches) {
>> // There are items.
>> // Do some other stuff.
>> }
>> }
>> 
>> // With suggestion:
>> foreach (string m; r.match(expr)) {
>> // There are items.
>> // Do some stuff.
>> } else {
>> // There are no items.
>> // Do some other stuff.
>> }
>> 
>> What do you guys think?
>> --AJG.
>> 
>> 
>> ==========================
>> if (!sin) throw (rock[0]);

I really like this.  I've found myself doing the if (length == 0) { } else foreach ... quite a bit.  It really would clean things up a bit.

Regards,
James Dunne

>I really like this.  I've found myself doing the if (length == 0) { } else foreach ... quite a bit.  It really would clean things up a bit.

Exactly. I'm glad somebody else finds it useful. I've done the (length == 0) roundaround about a million times in PHP, C# and now in D. It'd be great to have that kind of support in the language.

Cheers,
--AJG.

================================
2B || !2B, that is the question.