>>>>> "Peter" == Peter Eisentraut <[EMAIL PROTECTED]> writes:
Peter> Okay, so I don't claim to know a thing about cross-compilation,
Peter> but the current confusion around these parts doesn't help
Peter> either.
2 * likewise :)
Here is my understanding, plus some of my opinion (forged by listening
to the thread). Note that we are talking of a brand new situation,
the previous definitions are changed.
- Ideal cross compiling situation:
The user said --host. This gives a better control than build != host.
- Historical cross compiling situation:
You can't run what you compile.
The latter should be deprecated, because of all the problems it
implies (it hides failures from the native compiler etc.).
Unfortunately, since it will be years before all the `configure' use
the new Autoconf, we will have to maintain this for a long time (well,
I'm saying this because I'm not so strict with backward compatibility,
some might say `for the eternity' :).
But ideally, IMHO, the definition should be limited to the first item.
There will be a patch so that when given --host=foo, Autoconf will
look for foo-cc etc. so from the point of view of the comfort,
the new situation is a step towards people who have to cross compile.
On the other hand, because the historical definition hides problems,
as a step towards ``regular'' users, I'm tempted to have this kind of
cross compilation situation clearly displayed by a warning (something
like `can't run executables! Seems to me your compiler is cross
compiler. Next time, please use --host).
Akim
PS/ IANACCU, I Am Not A Cross Compiler User.
