NT client?

[Love]

Dr A V Le Blanc <LeBlanc at mcc.ac.uk> writes:

> assar at stacken.kth.se wrote:
> >The current code works in test mode (--test) under NT with
> >cygnus-win32.
> 
> Well, here is a report about this.  I found that I could get
> KTH Kerberos to compile, once I compiled gdbm and fiddled one
> statement which called the non-existent ruserok() function.
> (Although it compiles, kinit does not work; but this is more
> of a problem for KTH Kerberos than for arla.)

The problem is (if you are using B20.1) that gettimeofday() doesn't return
time of day (but rather 7 Jan 1970). Dummy implemation with time()
worked. (Its a known bug).


> I then tried to compile arlad.  In lib/roken I had to add an
> '#include <sys/socket.h>' to recmesg.c and to sendmesg.c.

Hmm, we didn't need this.

> In lib/ko/gensysname.c, an enum uses the constant O_TEXT,
> which already has a meaning from fcntl.h; I just redefined
> it to O_TTEXTT.  

Already fixed in the tree.

> Finally, in xfs/unknown/xfs/xfs_deb.h I
> had to comment out the #include line, which has nothing
> to include.

It should include xfs/include/xfs/xfs_debug.h.

> And in doing a 'make install', I had to
> edit the Makefiles in appl and in arlad, which had created
> zzz.exe, but were trying to install zzz for each of the programs.

*Sigh* don't know if I want to add that glue yet.

> At this point, I tried running arlad in test mode.  Despite
> not having a ticket, it did indeed see the root of /afs and
> allowed me to cd and to cat files.

*Wow* that is better then us, assar patched up rx to do more sane things,
and after that is worked. (recvmsg returned a strange errno that it
wasn't supposted to return but want the only sane thing to report
when converting it from Winsock).

Love