"VANCE,JEFF (HP-Cupertino,ex1)" wrote:
>
> Hi all,
>
> The sendmail Build script has this command in it:
>   $arch=`uname -m | sed -e 's/ //g' -e 's/\//-/g'`
>
> Echoing $arch on invent3k reveals:
>   SERIES989-600
>
> This string will be hard to map to MPE-iX as the OS
> profile file to use.  The Build script allows some
> tweaking on the value returned by uname, but is there
> a better way?  A $uname with no args produces "MPE/iX"
> and seems to be more useful here...

IIRC, sendmail uses the os for the build directory (this is from 8.9.1):

        os=`uname -s | sed -e 's/\//-/g' -e 's/ //g'`

Which generates:

        MPE-iX

I've just put my old 8.9.1 diffs online at:

        http://www.bixby.org/mark/sendmail-8.9.1-diff.txt

Use at your own risk.  Some of the stuff in there may no longer be needed, is
flat out wrong, or I would do differently if starting over.  Note that there is
some duplication in the diff because some files in the directory tree are
really symlinks to other files.

The stuff that is wrong is fsync() and ftruncate().  I only recently discovered
that the third parameter to FCONTROL() needs to be a long pointer, and the
sendmail code generates a short pointer when compiled with gcc.  I have correct
fsync() and ftruncate() code (that passes a long pointer to FCONTROL()) in
PostgreSQL and BIND 8.2.4 beta.
--
[log in to unmask]
Remainder of .sig suppressed to conserve scarce California electrons...

* To join/leave the list, search archives, change list settings, *
* etc., please visit http://raven.utc.edu/archives/hp3000-l.html *