arla rpms for redhat 6.0?

[Assar Westerlund]

"Brandon S. Allbery KF8NH" <allbery at kf8nh.apk.net> writes:
> Snippets from my KTH build script:

All these should problems should been fixed in the current snapshot (ftp://ftp.pdc.kth.se/pub/krb/src/alpha/krb4-19990508.tar.gz.

> # sddd: -lndbm isn't there, but -ldb contains it.  and isn't checked.

It now tests for db.h (and adds the magic DB_DBM_HSEARCH==1).

> # also need KTH's ndbm.h.Linux (which is now in build dir sans suffix)
> *_linux2[2-9]*)
>     CC="${CC-gcc} -I`pwd`" export CC
>     LIBS="$LIBS -ldb" export LIBS
>     conf="$conf --with-mailspool=/var/spool/mail"
>     cfghack=y
>     ;;
> 
> # ( . . . )
> 
>     ./configure --prefix="$prefix" --disable-cat-manpages $conf
>     if [ $cfghack = y ]; then
>         # glibc2.1 exports getmsg(), but Linux lacks STREAMS
>         ed - config.status <<\EOF
> g/HAVE_GETMSG/d
> g/STREAMSPTY/d
> w
> q
> EOF

There's also a test that catches the non-working getmsg.  streamspty
should not be turned on if there's no getmsg.

> There is one more issue:  seems the check for utmpx.h use of __e_termination
> and __e_exit doesn't work.  I added ifdefs based on glibc version to hack
> around that, but a fixed configure check is the correct answer (because it
> tries to detect that already; it fails with glibc2.1's include files for some
> reason).  This affects appl/bsd/rlogind.c and appl/telnet/telnetd/sys_term.c.

I think there are two problems here.  The utmpx tests were wrong (they
tested in utmp.h instead of utmpx.h) and you need to define
__GNU_SOURCE, both of which should be fixed.

> The resulting krb4 is running quite nicely on Red Hat 6

good

> aside from the fact that it doesn't understand the (optional) new
> MD5 passwords.

That's no surprise.  How are these passwords stored?

> (It looks like it might be compatible with 
> BSD MD5 passwords, so that could well solve two platform issues in a single
> stroke.

Yeah, sharing the same code for that part would be the best way of
doing it, IMHO.

/assar