Re: [boost] Behavior change in path operator /

Subject: Re: [boost] Behavior change in path operator /
From: Gérald Fauvelle (gf_at_[hidden])
Date: 2009-07-16 03:20:20

Le 15/07/2009 23:22, Scott McMurray a ecrit :
> 2009/7/15 Gérald Fauvelle <gf_at_[hidden]>:
>>
>> I've tried to find when, in what version,  this change occured, but couldn't
>> find it in release notes.
> 
> In the current version, at least, this behaviour for / comes from the
> explicitly-specified /=, so checking the docs for that should fine the
> change.  I don't know why it was made, though I suspect

"The path stored in rhs is appended to the stored path."

Well, that's clear, but it doesn't say why it changed, and why it can 
return a non valid path.

>> And if I want previous behavior, I have 2 solutions. Fix my code, of fix
>> boost code. I'm afraid that if I fix boost code I could break some boost
>> functionnalities, or can I do it with no problem ?
> 
> If you decide to fix your code, apparently the function that does what
> you want in the current version is named complete:
> http://www.boost.org/libs/filesystem/doc/reference.html#Operations-functions

Thanks, for the moment, I fixed some part of my code checking path with 
is_complete, but as we use this almost everywhere, I think I would 
prefer to do it in boost.

Are there unit tests of boost path that I can run to be sure that 
nothing is broken ?

Gérald