[Raw Msg Headers][Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: aliases problem



That's why I've been using a small patch to allow Smail3-style
"real-username" syntax instead of troublesome "\username" for many
years...  Somewhere around line 850 in proto/cf/aliases.cf:

# check for "real-userid" and deliver locally bypassing aliasing
       ssift "$user" in
       real-(.*) db add expansions "$key" local
               return (((local \1$plustail$domain \1$plustail$domain $attr))) ;;
       tfiss


On Fri, 2004-06-25 at 02:51 +0300, Matti Aarnio wrote:
> On Thu, Jun 24, 2004 at 04:34:36PM -0400, Peter Ip wrote:
> > I'm trying to get vacation working in aliases in 2.99.56. I followed the
> > instructions on the manpage and put in an entry like:
> > 
> > peter:	"\peter", "/usr/local/lib/mail/vacation peter"
> > 
> > The router returns:
> > 
> > error nosuchuser ter g0
> > 
> > If I leave out the backslash, I get
> > 
> > bitbucket 'duplicate-"peter"-aliases' peter g1
> > 
> > It works if the entry is in .forward.
> 
> Umm...   The back-slash processing is troublesome, to say the least..
> 
> It works with  .forward,  because  aliases.cf  treats its expansion
> wrapping with "been here already, skip to next mechanism",  but with
> aliases the treatment is: "been here already, we are looping!"
> 
> It stems also from thinking of:
>    "user can do foolish things, sysadmin won't"
> Perhaps the definition of "foolish" needs to be adjusted..
> 
> 
> Testing the alias-expanded processing, I see that:
> 
>        dequote -- "\b-test"
>        tmp='\-test'
> 
> which is ...  suboptimal.
> 
> The more I think of it, the more I think that there must not
> be any backslash-stripping in the dequote processing, but
> complimentarily there needs to be explicite leading backslash
> stripper somewhere in the  aliases.cf's   routeuser()  code.
> 
> Err...   There is:
> 
> 
> 	a="$(dequote -- "$user")"
> 
> 	ssift "$a" in
> 	\\(.*)	user="\1" # Back-quoted username -- most likely
> 		#didexpand=local
> 		;;
> 	...................
> 	tfiss
> 
> Converting that dequote-call to explicite patterns in ssift is
> fairly easy, but ...  it is way too late tonight; must think more
> of that latter in upcoming weekend..  (but at first, finnish
> mid-summer rites are to be performed -- popular festivities
> in this country..)
> 
> > TIA
> > 
> > Peter
> > _____________________________________________________
> > Peter Ip, PhD
> > Computing and Network Services, University of Toronto
> > email: peter.ip@utoronto.ca
> 

This is a digitally signed message part