"Jeffrey W. Baker" wrote:
> Is it actually possible to compile sane? I downloaded both parts of the
> 1.0.4 package. I tried to build the backends, but sanei_scsi.c has some
> preprocessor horkage by which it includes /usr/include/scsi/sg.h and
Hmmm. Including both of those files is asking for
trouble. Even though I included a C++ type guard
inside sg.h , the GNU library maintainers changed
it in their version! Even then, the "guard" solution
is less than desirable because it just picks
the first header seen.
> which are clearly the same file and gcc
> gets pissed off by redefinition. After I fixed this problem (not the
> right way, I'm sure, but by hard-coding #include <scsi/sg.h>), the package
> compiled and installed.
It may have compiled and worked, but it defeated
what sanei_scsi was trying to do.
sg.h is collateral damage in a minor GNU-linux turf
war. The file /usr/src/linux/include/scsi/sg.h is the
header that matches the the sg driver in the kernel
(unless the kernel source and the running kernel are
out of sync). The file /usr/include/scsi/sg.h "belongs"
to the GNU library and may or may not match the running
sg driver. Surprisingly the glibc 2.2 library (update to
RH 7.0) does have the latest sg.h (correct for lk 2.4.0),
albeit reformatted and with a different guard. Perhaps
I should change the kernel guard to be the same as the
GNU guard ...
It seems bizarre to me that if the sg.h lived in the
"linux" directory rather than the "scsi" directory,
this problem wouldn't happen (since the former directory
is a symlink to /usr/src/linux/include/linux). There
seems little I can do about this other than try and
explain what is happening. The GNU library maintainers
have another spin on this, quoting Linus Torvalds
amongst others that apps should never directly access
Thanks for letting me share your infuriation ;-)
BTW SANE 1.0.4 built "out of the box" for me.
-- Source code, list archive, and docs: http://www.mostang.com/sane/ To unsubscribe: echo unsubscribe sane-devel | mail firstname.lastname@example.org
This archive was generated by hypermail 2b29 : Fri Jan 12 2001 - 10:51:12 PST