[RFC] Do we care about binary compatibility of code produced by cross-compilers?

[Paolo Carlini]

Ian Lance Taylor wrote:
> I don't think it's worth worrying about the fact that the tests are
> run multiple times.  If you want to speed up configury, help get Tom's
> quagmire project working for gcc (http://code.google.com/p/quagmire/).
>   
Interesting, I'll have a look.

To be clear, I was not talking *performance*. I was surprised that we 
were running exactly the same test 4 times, with the same outcome, of 
course. Sure, if those libraries are meant to be built separately, 
aren't really meant to be just "helpers" for GCC, that makes a lot of 
sense. But is that really true? For what is worth, as regards 
libstdc++-v, which I can say to know, history included, I can say we 
definitely moved away from the idea of delivering a self-consistent, 
closed, library. We are totally relying on GCC as C++ compiler, not just 
that, the exact GCC built in the tree. Thus, as far as v3 is concerned, 
it would be totally ok to "get from somewhere" the outcome of 
GCC_CHECK_TLS and just use it.

Paolo.