Jonathan McDowell <[EMAIL PROTECTED]> writes:

> Running ./configure I get:
> 
> checking for BSD pty names... sort: read error: Bad file descriptor
> done

I seem to recall a bug in some version of /bin/sh. Can you try running

  bash ./configure

and see if that works better?

> Then a 'make':
> 
> unix_user.c:393: structure has no member named `ut_user'

Seems I missed that usage of ut_user. I'm checking in this patch:

  diff -u -a -r1.14 unix_user.c
  --- unix_user.c 2000/05/08 20:17:26     1.14
  +++ unix_user.c 2000/05/09 08:29:18
  @@ -390,7 +390,12 @@
         if (log)
            /* FIXME: It should be safe to perform a blocking reverse dns lookup here,
             * as we have forked. */
  +#if HAVE_UT_NAME
  +         logwtmp(log->ut_line, log->ut_name, log->ut_host);
  +#elif HAVE_UT_USER
            logwtmp(log->ut_line, log->ut_user, log->ut_host);
  +#endif
  +
   #endif /* WITH_UTMP && HAVE_LOGWTMP */
  
         if (getuid() != user->super.uid)
  
> <utmp.h> needs <sys/types.h>; could be causing problems in the configure
> script?

To me, the configure problem seems unrelated to the tests for utmp
members. If you got HAVE_UT_* defined properly, that part of the
configure script seems to work.

> If I define PTY_BSD_SCHEME_FIRST_CHARS and PTY_BSD_SCHEME_SECOND_CHARS
> in config.h then I get listed in 'last' but not 'w' (so that's wtmp
> stuff working, but not utmp, right?) and I get a prompt ok.

Correct. That's as good as it gets with the current version. 

> Getting there. I'll see if I can figure out why the configure script
> doesn't grab the PTY stuff properly and also poke utmp.

Better utmp support would be great. Please send patches (preferably in
unified format, as produced by diff -u) to the list, as I will be
unavailable for some time.

Happy hacking,
/Niels

Reply via email to