Subject: Re: CVS commit: pkgsrc/mail/mimedefang
To: Mark Davies <mark@mcs.vuw.ac.nz>
From: John Nemeth <jnemeth@victoria.tc.ca>
List: netbsd-users
Date: 07/22/2003 05:28:25
On Nov 28, 11:36pm, John Nemeth wrote:
} On Nov 23,  3:46pm, John Nemeth wrote:
} } On Nov 24,  6:54am, Mark Davies wrote:
} } } 	From:  jnemeth@victoria.tc.ca (John Nemeth)
} } } 	Date:  Thu, 3 Jul 2003 16:07:59 -0700
} 
}      Hmm..., talking to myself...

     Hmm...  talking to myself some more...

} } } >      Do you have MIMEDefang running?  If so, on which version.  I tried
} } } > running it on 1.5.2 and didn't have any luck.  So, I upgraded to 1.6.1
} } } > and still didn't have any luck.  All it does for me is crash when
} } } > sendmail tries to connect to it.  I have perl-5.8.0nb1 and pth-2.0.0
} } } > installed.
} } } 
} } } Well our mail hub is actually a Solaris box, so I know the package works 
} } } there.  Unfortunately for this package the extent of the testing I can do on 
} } } NetBSD is that it builds (actually thats not true, given a little time I could 
} }      Building it isn't a problem.  Running it is a different story.
} } 
} } } rig up something to test it out but its not my standard setup).
} } 
} }      I've been thinking of rigging a test on a -current system to see
} } if maybe the problem is with pth.  If it works on -current, maybe I'll
} } try different pthread libraries on the 1.6.1 box.  I'll report back
} } after I do some more tests.
} 
}      I tried it on a -current system usings sources from July 5th.  It
} worked (sans a bug in the default filter).  I guess next will be to try
} a different pthread library on 1.6.1.

     I tried it on 1.6.1 using all four versions of the pthread library
that were available in pkgsrc.  The pthread library that is "blessed"
is GNU pth.  When I used that, the slaves would crash when sendmail
would try to connect to them.  The next one I tried was
unproven-pthreads.  With this one, the multiplexor found the slave busy
when sendmail tried to connect, so it would kill the slave and return a
tempfail error to sendmail.  The other two, mit-pthreads and ptl2,
we're too old.  Both of them included ancient versions of resolv.h
which prevented libmilter from building.  I might have been able to
make libmilter build, but it would have been a lot of work.  At this
point, as far as I'm concerned, MIMEDefang does not run on NetBSD
1.6.1, unless somebody can demonstrate to me otherwise.

}-- End of excerpt from John Nemeth