On Fri, 29 Mar 2002, [iso-8859-2] £ukasz Grochal wrote:
> You mean forcing reply to be temporary error regardless of what
> local_scan() really returns? I honestly don't see a reason for doing
> that. It should be responsibility of the person who writes the
> local_scan() function to ensure that it will finish in a reasonable
> time and decide, if there should be 2xx, 4xx or 5xx if it doesn't.
Indeed it should, but also Exim should perhaps protect itself...
With a timeout, local_scan() won't get a chance to return. It will get
chopped.
> That however reminds me of what I miss the most in the local_scan API
> - configurability. For now, the only way I see to pass configuration
> parameters (let alone debug level) to the local_scan() is to make it
> use its own configuration file. That means two configs - one for exim
> and a separate one for a function compiled in it. I wonder if it would
> be possible to somehow register some custom configuration variables
> (name, type perhaps, and default value) that one would then put into
> the main section of exim config and to refer to them somehow during
> runtime. That would make writing a customizable virus scanners based
> on local_scan() much easier.
I'll think about that one. It might be possible to have a "local scan"
part of the configuration, which would allow you to define things
within your own local_scan code module.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
