Re: [boost] [Serialization] why operator &()?

From: Robert Ramey (ramey_at_[hidden])
Date: 2005-10-14 14:15:21

Jim Hyslop wrote:
> Robert Ramey <ramey <at> rrsd.com> writes:

> Templates are a red herring here - the use of operator & in this case
> is independent of templates.

Maybe - but its not independent of the type.  The key feature of the & 
operator in this instance is that
its meaning save vs. load changes depending on the type its applied to. 
Generally I dislike and avoid this kind of invisible mode shifting.  But in 
this case it has proved more useful than error-prone.

>>  I suspected that the & operator
>> was chosen due to its operator precedence being at a particular spot
>> in the operator precedence hierarchy[...]

> Except that operator precedence only applies to "built-in" operators
> - user defined operators are function calls, and have that level of
> precedence.

I realize this.  But once one decides he wants to overload an operator, one 
has to pick an operator with consideration of its precedence.  So that's the 
main reason I stuck with it.

>> The usage of << and >> are my own and are obviously analog to the
>> concept of using << and >> for stream input / output respectively.

> These I understand and agree with, because the analogy is well-known
> and understood.

Whow knows, maybe someday & will be added to the above.

>> So, for better or worse, that's how we arrived here.

> Thanks for the explanation. Now that the library has been accepted,
> I'm not expecting this will change (at least not in the near future),
> I just wanted to know the rationale behind it. I'll have to pay more
> attention in the future ;=)

I don't think coming in earlier would have changed this.  The usage of the & 
operator for this purpose seems to have been pretty well recieved.  And of 
course it's not obligatory.  One who believes its a detriment can just avoid 
its usage.  That is why is_saving and is_loading are include in the archive 
interface.

Robert Ramey