acbhour@accucx.cc.ruu.nl (Rudi van Houten) (10/19/90)
If a message is queued by sendmail, as is all incoming UUCP mail handled by /bin/rmail, the recipient's address is evaluated from the user's .forward. When sendmail then executes the filter-program in the queue run the program is not run with the correct uid (since sendmail doesn't know that) and filter is unable to find the ~/.elm/filter-rules. Result: the message is noiseless placed in daemon's mailbox. I see two solutions to fix the problem: 1.simply add a obligate parameter to filter, being the concerning username. This is very simple to implement, but it changes the user's doc. 2.teach sendmail not to evaluate the .forward-file if it expands to a prog-mailer (starts with the pipe-symbol). Has someone a better idea?
andy@jhunix.HCF.JHU.EDU (Andy S Poling) (10/22/90)
In article <870@accucx.cc.ruu.nl> acbhour@accucx.cc.ruu.nl (Rudi van Houten) writes: >If a message is queued by sendmail, as is all incoming UUCP mail handled >by /bin/rmail, the recipient's address is evaluated from the user's .forward. >When sendmail then executes the filter-program in the queue run the program >is not run with the correct uid (since sendmail doesn't know that) and >filter is unable to find the ~/.elm/filter-rules. Result: the message is >noiseless placed in daemon's mailbox. >I see two solutions to fix the problem: >1.simply add a obligate parameter to filter, being the concerning username. > This is very simple to implement, but it changes the user's doc. >2.teach sendmail not to evaluate the .forward-file if it expands to a > prog-mailer (starts with the pipe-symbol). > >Has someone a better idea? I ran into this a long time ago with 5.61 and the problem remains unsolved at least in 5.64 . We queue almost all of our mail messages here and run the queue very often - this way, sendmail always returns a zero exit-status to whatever invoked it. I chose an even simpler solution thatn Rudi's second one and posted a simple (about 8 lines of code) patch to alias.c which causes evaluation of the .forward file to always be delayed if we will be queueing the message. I can mail the patch to anyone who missed it. -Amdy -- Andy Poling Internet: andy@gollum.hcf.jhu.edu UNIX Systems Programmer Bitnet: ANDY@JHUNIX Homewood Academic Computing Voice: (301)338-8096 Johns Hopkins University UUCP: uunet!mimsy!aplcen!jhunix!andy