Vin Shelton <acs(a)xemacs.org> writes:
Jerry James <james(a)xemacs.org> writes:
> Vin Shelton <acs(a)xemacs.org> wrote:
>> It may not be your compiler.h patches, Jerry - it's been about 10 days
>> since I've built 21.5 on this platform, but I can no longer build on
>> my linux box with this config:
>
> No, that's due to yet another patch of mine, "Module extension
> rationalization". I am clearly trying to make your life as
> difficult as possible, Vin. :-( Did you configure --without-modules?
> It looks like you must have, and the logic for handling module
> auto-autoloads (i.e., ignoring them) is failing in that case. I'll
> get it fixed in a jiffy.
It was not my intention to build --without-modules, and in fact, I did
not do so explicitly.
FWIW, my gcc build did work, and it *did* include module support. The
icc build did not include module support, although it used the same
configure options (modulo choice of compiler and CFLAGS).
I'll try to investigate a little further.
I now see that all my icc builds (both 21.4 and 21.5) fail to detect
that I can build modules:
checking for module support
checking for dlfcn.h... yes
checking for dlopen in -lc... Defining HAVE_DLOPEN
checking how to build dynamic libraries for i686-pc-linux
checking checking whether we are using GNU C... no
checking how to produce PIC code... none
No module support.
Hmmm.
- Vin