Re: [patch] : Add more symantics to sort, partial_sort, nth_element

[chris jefferson <caj at cs dot york dot ac dot uk>]

Paolo Carlini wrote:

> chris jefferson wrote:
>
>  
>
> > Exactly what it says in the title. This is the non-contraversal part
> > of the patch I sent last week. I'm still tidying up the other half.
> >    
>
> Ok, I'm going to apply to v7 the patch basically as-is, besides a few
> trivial reformattings and missing std:: qualifications: passes
> regtesting, and looks basically sane. Also, v7 is for experimental code
> and we can afford a little bit of risk, for now. I will study the code
> in better detail over the next days, anyway.
>
>  
(Famous last words) I feel fairly confident about all the code from an 
algorithmic viewpoint (I've stepped through it all in a debugger), 
although it's not totally impossible (I'm fairly confident however) that 
I haven't accidentally made a slight assumption I'm not permitted to 
use. I've been finding it fairly hard to find big pieces of C++ code 
which really push the standard library to be sure.

Sorry about the qualifications...

> However, a *necessary precondition* for even envisaging merging to
> mainline this code, is documentation on top of the new helper functions
> (like __introsort_partition for moveable, __unguarded_*). I know that
> comments are also lacking on top of some HP/SGI functions but we don't
> want to follow forever those bad examples ;) We are risking maintainance
> nightmares, otherwise.
>
>  
Yes, I should have documented that.. sorry. I'm going to take the time 
to step back and do a bit of cleaning now I think, before submitting 
stable_sort / inplace_partition (which are quite intrusive).

> Also, a trivial patch replacing everywhere __is_moveable::value with
> __value is *very* welcome: user code *can* define a value macro, remember.
>
>  
It looks like I don't have to now :) But sorry, I should still have done 
this some time ago!

Chris