[boost] Re: boost::tuple to MPL sequence

From: Alexander Nasonov (alnsn-mycop_at_[hidden])
Date: 2003-04-29 09:48:13

David Abrahams wrote:

> Alexander Nasonov <alnsn-mycop_at_[hidden]> writes:
> 
>> David Abrahams wrote:
>>> How does that help with your own sequence?  It seems to me that (given
>>> a conforming compiler), the concept docs should be sufficient.
> 
> I'm not saying the docs are as complete as they should be, but...
> 
>> I see this in code but not in the documentation:
>> 1. Sequence tag
> 
> You don't need sequence tags if you have partial specialization.  It
> *is* convenient, though.

Does it mean that it's better to specialize clear rather then clear_traits?

>> 2. You can specialize erase_traits or erase
> 
> You can always specialize.  Though I guess it's of marginal morality
> to do that on built-in types like functions.
> 
>> 3. Default version of erase uses clear and push_front => it's enough to
>> specialize push_front_traits and clear_traits if you want erase for
>> your-own sequence
> 
> Yes, that's another _convenience_ you can't take advantage of without
> better docs.
> 
>> If it's just an implementation detail I give up but I would like
>> ExtensibleSequence concept to be open.
> 
> I don't understand what you're asking for here.
> Would you like to propose a patch for the MPL docs?

I'm just asking for better docs.

>>> Would you like to propose a patch for the current CVS tuples?
>>
>> Aleksey almost did it:
>> http://lists.boost.org/MailArchives/boost/msg46780.php
> 
> Would you like to make it complete?
I can try when I finish my current work.

> The MPL docs should be better than they are.  They're getting there,
> slowly, but it would sure help if people who have discovered the inner
> secrets would contribute them back in appropriate form.

I can try to write How_to_implement_extensible_sequence.html.

-- 
Alexander Nasonov
Remove minus and all between minus and at from my e-mail for timely response