Re: 2nd: [boost] shared_ptr proposal

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-07-13 11:43:46

At 10:37 AM 7/13/2001, Peter Dimov wrote:

 >> They are currently unspecified.  At least for now, that's probably
 >> best.  See below.
 >
 >IOW, a linked-list implementation is conforming. This is good. :-)

Yes.

Complexity requirements can help ensure that application programs ported to 
other library implementations still perform reasonably.

But there can be a real downside.  They discourage implementors from 
innovating.  They are hard to specify correctly.  They effectively add to 
the cost of standardization.  So they should only be used where there is a 
major advantage to doing so.

 >
 >>  >An interesting related question is "why does shared_ptr have
 >use_count?"
 >>
 >> As an aid to writing test cases and debugging displays.  One of the
 >> progenitors had use_count(), and it was useful in tracking down bugs in 
a
 >> complex project that turned out to have cyclic-dependencies.
 >
 >This should probably be documented (along with the decision to not 
specify
 >complexity requirements.)

Done.

 >Why is this important? Because some nice interface additions to 
shared_ptr
 >(my shared_ptr_id proposal, for instance) would be inappropriate under a
 >linked-list implementation.
 >
 >Perhaps this deserves a new FAQ entry:
 >
 >Q: Why doesn't shared_ptr provide (your pet feature here)?
 >A: Because (your pet feature here) would mandate a reference counted
 >implementation. This is not the intent.

Added.  Thanks!  And thanks to Christian for posing the questions in the 
first place.

--Beman