This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Make/bootstrap failure on i386-apple-darwin8.10.1 (Intel Core Duo)

    Actually the usage of --host in the fink gcc packaging predates
my ownership of it. I believe the fact that Apple configures their
gcc builds with...

 --build=powerpc-apple-darwin8 --host=powerpc-apple-darwin8 --target=powerpc-apple-darwin8

...may have been the motivator. This actually will be trival to fix since we will
just change the egrep command we use to parse out the --host field from the output
of 'cc -v' to parse out the --build field instead. On Macintel, I think the usage
of --host was also meant to insure that gcc was built for i686-apple-darwin8 instead
of i386-apple-darwin8.

On Thu, Oct 04, 2007 at 03:25:45PM +0100, Dave Korn wrote:
>   Could you just repeat that once more, slowly, for those of us who are a bit
> hard-of-thinking?  I'm not quite clear about how, if you're building a normal
> package, i.e. one that runs on the build machine, why you'd need to tell it
> "which host to build it for" at all, since the answer, by definition, is "this
> one".
>   I always thought that if you want to build a "normal" package, i.e. one that
> "will run on the machine on which you are doing the build" - what I've always
> referred to as "a native build", you simply don't specify anything at all
> (well, don't specify any of the --{build,host,target} trio, to be exact).
>   But yeh, I get the part about how it now assumes that if you're specifying a
> --host, you must be doing a cross-build, because if --host was the same as
> --build, you wouldn't specify either ...
>     cheers,
>       DaveK
> -- 
> Can't think of a witty .sigline today....

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]