A comma in an if or while

[mcleano]

That title is odd i know but I just couldn't think of a short and easy way to sum it up. Basically, I remember reading on one of the forum threads that if you had for example: if(/*statement 1*/, /*statement 2*/); that statement 1 would always be true and would always be executed. Firstly, is this true? And if so, what are peoples views on using this?

This is an example of code that I have use this method with:

if(getline(std::cin, strInput), !(strInput.compare("")))
	continue;

[moorecm]

This reminds me of this: http://www.cplusplus.com/forum/beginner/22678/

If I remember correctly, the sub-statements will all be executed but only the last one (on the right) will be used for the condition.

As far as my view on when its use is appropriate, I vote for never.

[Albatross]

I vote for never, and I hope that when C++0x comes out, it's not just deprecated, it's removed.

Okay, it has its uses.

-Albatross

[Bazzy]

It's the purpose of the comma operator, it executes the left operand and returns the right one
It's quite a good feature. The problem is that some people read it as 'and' or something like that
It shouldn't be removed or deprecated

http://www.cplusplus.com/doc/tutorial/operators/

[Albatross]

All right, the operator has its uses in the contexts listed.

I still vote for never. In my opinion, we should limit the number of calculations we do per inter-semicolon space. However, that comma in that context is basically just a way of fitting as many of those calculations as possible into a certain line. And when you have an if attached to that...

-Albatross

[mcleano]

Thanks for your replies people!

[chrisname]

I'm with Albatross. Although it has it's uses, it looks very abusable. Still it shouldn't be deprecated or removed.

[mcleano]

@chrisname ... With regards to my example, what are you're thoughts?

[guestgulkan]

If the standards committee took away a C++ feature that at least one member of the forum was uncomfortable with /or afraid to use, there would be fu*kall left of the language.

[helios]

Removing that operator would break this for (/*...*/;a++,b--) //... and this #define M(x) (x+=1, x*2) .
It's very easy to update a language when you don't have to deal with minutia, such as backwards compatibility. The problem is that you do have to, and removing operators is one of those things that break it.

Hey, since we're removing operators, why not merge assignment and equality into a single operator and add <> as inequality? Surely nothing can go wrong there!