[HH] c++ strings?

[Greg London]

So, if I have a subroutine that passes by reference,
and I want to pass in something that can't be modified,
I have to put it into some intermediate variable first?

That seems.... inefficient.



> On 11/20/2012 03:57 PM, Matthew Gillen wrote:
>> On 11/20/2012 03:41 PM, Greg London wrote:
>>> Why would I need "const" when I pass by reference, but not need const
>>> when
>>> I pass by value?
>>
>> Because references are not pointers, and you cannot "re-seat" the
>> reference.
>>    http://www.parashift.com/c++-faq/reseating-refs.html
>>
>> If you want the nitty-gritty details on how references are implemented:
>>   http://www.parashift.com/c++-faq/overview-refs.html
>
> Sorry, I jumped the gun.  Your problem had nothing to do with re-seating
> a reference.
>
> The problem in your example:
>   ...
>   Note("hello world\n");
>   ...
>   void Note( string &msg){ ...
>
> is that "hello world", by virtue of being a string literal, *is* a const
> string, but your function prototype (when sans-const) is indicating that
> the function reserves the right to modify the string.
>
> When you pass by value, you're always getting a copy, so the effect is
> essentially:
>   Note(new std::string("hello world\n");
>
> ...and there's never a problem modifying your local copy of the original
> string. However, if you pass by reference, you're not making a new copy,
> so if the source object has some restrictions (e.g. is const), then you
> can't pass it to functions that reserve the right to modify it.
>
> Const-correctness can be tricky when you're first learning it, but if
> you get it right it is a great tool to make the the compiler help you.
>
> Matt
> _______________________________________________
> Hardwarehacking mailing list
> Hardwarehacking at blu.org
> http://lists.blu.org/mailman/listinfo/hardwarehacking
>


--