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

[Paolo Carlini]

Hi again,

> That said, I think Ian's right; it's better to accept the ABI
> breakage than to have cross- and native-compilation differ.  The latter
> problem is just too horrid.

Ok, agreed. Let's try to make an effort in this direction for 4.4.0, I think we have time to make good progress.

The most important point in needs of clarification is the status of the TRY_LINK tests, otherwise we have only a couple of TRY_RUN currently (of minor importance) and I'm confident we can (we are already) hard-wiring satisfactorily otherwise, maybe we'll need the feedback of the target maintainers to complete a few hard-wirings...

Paolo.