C++ Logo

SG12

Advanced search

Subject: Re: [ub] Justification for < not being a total order on pointers?
From: Gabriel Dos Reis (gdr_at_[hidden])
Date: 2013-10-10 20:37:30


Nevin Liber <nevin_at_[hidden]> writes:

| On 10 October 2013 18:24, Gabriel Dos Reis <gdr_at_[hidden]> wrote:
|
| Nevin Liber <nevin_at_[hidden]> writes:
|
| | [Y'know, I really wish replies would by default go the whole ub group...]
|
| I cannot control your email software, but when you sent me the message,
| it wasn't addressed to the group and I replied the same way.
|
|
| It isn't my software (Gmail web client); it's the mailing list.  All the other
| reflectors default to replying to the group; this one defaults to replying only
| to the original author.  I'm not sure who to contact to get that changed.
|  

I believe the setting is the same as for all the other SG mailing lists
that run on isocpp.open-std.org. I will check.

[ when I receive I message not sent to the list I always treat it as
  a concious decision from the sender as private message, otherwise
  he or she would have hit the "reply all" button (which I believe is
  also available in Gmail. ]

| As observed by Sean Parent in another discussion, std::less<T*> as a
| total order was never meant to be synonymous for operator< on pointers.
|
|
| Sean also says in [c++std-lib-ext-550]: The definition of std::less should be
| "a representational order if no natural total order (operator <) is available,
| otherwise the natural total order".  He goes on: "They are named separately, 
| operator<() is the natural total order and std::less<> is the natural total
| order if one exists, otherwise it is a representational order."
|
| Pointers clearly violate those rules.

Sean will speak for himself, but I don't think he was making the point
that operator< on pointers was a total order. My reading of his message
was the opposite.

| Of course, C++ experts live with these contradictory rules all the time.

I think there is a debate about what is perceived as contradictory.

|  That's what makes them experts... :-(

I must confess I do not understand why saying that for "p < q" to make
sense, the addresses must be related is an expert-level thing.

Certainly, we don't claim that for two random iterators p and q, the
expression p < q must make sense whether p and q are related. Are we?

I've not seen novices wanting to do anything sensible in a conditional
based on p < q when the addresses are not related. I have seen lot of
examples by experts, though. Most novices I've met or delt with come
with a background where the notion of pointer comparison does not even
exist to start with...

|  Some experts even argue we should keep this contradictory behavior
| without providing concrete reasons why it is still necessary.

Do you have links to such statements?

-- Gaby


SG12 list run by herb.sutter at gmail.com