Re: [v3] Fix / clean-up config vs crosses (2/n)

[Paolo Carlini <paolo dot carlini at oracle dot com>]

Mark Mitchell wrote:
> Actually, I don't know much about them, except from a very high-level 
> perspective.  I could go dive into it, of course, but for me this is 
> more theory than practice, I'm afraid.  There are so many systems out 
> there that I just think it pays to be paranoid.
>
> Concretely, I have no evidence of a system which puts writev in 
> headers without guaranteeing it appears in the C library, but I can 
> imagine it.
Sure. On the other hand, I'm not sure it pays to be paranoid ;) I mean, 
AFAIK, libstdc++ doesn't work well on many embedded systems for many 
other reasons besides this one. If we can be reasonably sure that on the 
widespread targets the issue with declared-not-used doesn't exist for 
some functions, we could, in my opinion, complete a nice clean-up of 
these cross-configury things, and leave special needs to command-line 
options - as you mentioned - or hard coded values, which would be 
implemented and contributed by the target maintainers actually caring 
about those systems on a case by case basis. That last point leads me to 
a related observation: often, unfortunately, maintainers of targets 
potentially benefiting from some changes do not exist, or do not care; 
in that case, sheer paranoia without any serious deep  knowledge of the 
targets doesn't seem the right attitude for a general maintainer of the 
library, in particular when it blocks improvements which would clearly 
benefit a large number of users, whose targets are well known and 
maintained.

Well, I have also a polemic point for people (not you, Mark, to be 100% 
clear!) using too easily the word "broken" (which, literally translated 
in italian sound really bad and impolite, maybe in English is different) 
without a constructive proposal and without clearly showing 
understanding of the existing trade-offs, but I will not insist...

Paolo.