>>>>> "Ossama" == Ossama Othman <[EMAIL PROTECTED]> writes:
Ossama> BTW, my comments below are more design philosophy related in
Ossama> nature than actual proposals for a change.
Nah, you don't need any excuse for expressing yourself :)
>> I'm not sure to understand what you are referring to. The patch I
>> sent sets cross_compiling to yes when --host. Later it checks if
>> the compiler is believed to be a cross-compiler (the usual
>> ac_prog_cc_cross etc.). If there is a mismatch (yes-no and
>> no-yes), it WARN complains, but will always proceed with a `yes'.
Ossama> Ah okay. I thought that the patch prevented this from
Ossama> occurring since you sent me a heads up about the patch.
Ossama> Sorry, I misunderstood. I seem to be doing that alot lately.
Ossama> :-(
I did so because if we want to proceed, it seems the most logical
choice, the one which will trigger more warning in the bad case, and
will work the most in the good cases. But really, I would prefer a
hard error.
Also, as Paul noted elsewhere (plus some of my own interpretation),
when --host is given, we shouldn't even run the cross-compilation test
for the compilers, nor AC_PROG_CC_WORKS etc.
>> No, I don't think so (but I confess I'm not in this situation, so
>> it's a pure guess I'm making), because --host is just what is
>> needed, and it is not that expensive.
Ossama> Fair enough. My cross-compiler aware projects document how to
Ossama> do a cross compiled build, and I expect to my users to RTFM.
Ossama> What you say makes sense.
I think we're on the right track. The big question is whether `--host
== cross-compilation' and `!cross-compilation && can't run executables
=> error' will be accepted. Ian, I'd like to have your opinion on
this.
>> Maybe I'm becoming a fanatic of `--host', but it seems to me that
>> you two people can perfectly agree on the definition
>> `cross-compiling == --host'. Paul is happy because by default
>> bizarre situations are denounced, and Ossama is too because it is
>> simple and clear. And remember this is valid for *any* package,
>> not only those with AC_CANONICAL_SOMETHING.
Ossama> Right. That sounds good.
Ossama> Thanks! -Ossama
Akim
