[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Problem with configure falsely detecting presence ofsystemroutines(w
Kelly F. Hickel
RE: Problem with configure falsely detecting presence ofsystemroutines(while porting to NSK)
Fri, 3 Oct 2003 16:32:43 -0500
> -----Original Message-----
> From: Derek Robert Price [mailto:address@hidden
> Sent: Friday, October 03, 2003 2:12 PM
> To: Kelly F. Hickel
> Cc: address@hidden
> Subject: Re: Problem with configure falsely detecting presence
> ofsystemroutines(while porting to NSK)
> It might be enough to say, "hey, configuring on this system fails
> because the linker doesn't report errors correctly and Floss deals
> it like this! Would you find this useful?"
> | [Kelly Hickel] As far as the _XOPEN_SOURCE_EXTENDED stuff goes, my
> |googling shows that it's present on several systems (including hpux,
> |z-os, tru-64). On the NSK it was needed to resolve some socket
> |It may be needed for other things, but it was definitely needed for
> Could you write a configure test that can determine that it is
> to define _XOPEN_SOURCE_EXTENDED? It is possible that the alternative
> is simply putting a note in INSTALL that it was necessary to supply
> option on Tandem. I have a dim recollection of a discussion on
> bug-gnulib about defining _XOPEN_SOURCE causing problems due to a
> difference of interpretation between platforms. I don't recall if the
> bugs were gross or subtle. Not sure what _XOPEN_SOURCE's relationship
> to _XOPEN_SOURCE_EXTENDED is either.
It's been a really long time since I wrote a conftest, long enough that
I'm not sure I've ever done it. I did whip up a little .c file to
demonstrate it, is this useful???
/tmp: cat test_af_inet.c
int main(int argc, char *argv)
/tmp: c89 -D_XOPEN_SOURCE_EXTENDED=1 -o test_af_inet test_af_inet.c
/tmp: c89 -o test_af_inet test_af_inet.c
"/tmp/test_af_inet.c", line 5: error(114): identifier "AF_INET" is
1 error detected in the compilation of "test_af_inet.c".
c89: /usr/lib/cfe exited, returning 2.