OT: SCO 6.0 - MAJOR BUG!!!!!

Jean-Pierre A. Radley appl at jpr.com
Mon Oct 24 08:01:23 PDT 2005


Jean-Pierre A. Radley propounded (on Sun, Oct 23, 2005 at 12:45:21PM -0400):
| Jean-Pierre A. Radley propounded (on Sat, Oct 22, 2005 at 07:26:33PM -0400):
| | John Esak propounded (on Sat, Oct 22, 2005 at 06:14:36PM -0400):
| | | Here is the old scenario. Imagine you are *running* rclerk. As the root user
| | | you go to some other session and try and do:
| | | 
| | |     cp /tmp/fred /u/appl/fp/rclerk
| | | 
| | | You will get a message something like cP can't do this for you because the
| | | text file is busy... etc., etc. This is fine and I think it should work this
| | | way.  On the new 6.0, if you do the very same thing (and imagine /tmp/fred
| | | to be a 6 byte file that holds the word "hello") what will happen is exactly
| | | this. You get back your prompt as if the /tmp/fred file *has* been copied
| | | over the top of the *running* rclerk file.  You know why? BECAUSE IT
| | | HAS!!!!!
| | | 
| | | The very next keystroke from the person who is *running* the rclerk
| | | generates a core dump!  This happens with any program and copying over it,
| | | etc. It is just unbelievable. If there were 20 people running rclerk, 20
| | | people each get core dumps on their next keystroke. This is just simply
| | | stupid, and in my opinion is a severe bug in the way things work.
| | | 
| | | Okay, I'm ready to be told why this is some kind of a good idea and not a
| | | pure giant bug.
| | 
| | It's a major bug, and I've posted your report to the people who can fix it.
| 
| The first answer I got is that is the usual behavior in Unixware.
| 
| That leaves open the question of whether it is or is not a bug.

A further response:

        FYI, this change goes back to System V Release 4, released
        in 1989.  It is a consequence of the adoption of SunOS VM
        architecture in SVR4 to replace the regions VM of earlier System
        V releases.
-- 
JP


More information about the Filepro-list mailing list