Re: [exim] linefeeds in check_data acl

Marten Lehmann wrote:

> Hello,
>
>> looking at spool_mbox.c, I'd say no (apart from changing the code).
>
>
> Too bad.
>
>> But why would you want to do that?
>
>
> Well, as I explained in my first email, there are different virus
> scanners and spamfilters that just don't work with solely LF-sequences.
> There isn't only ClamAV and SpamAssasin which maybe work fine with LFs
> (I haven't used them so far). So when the first task of someone using
> these scanners is to write their own function to map LF back to CRLF,
> why shouldn't exim include this so integrators can concentrate on more
> important parts?
>
> Regards
> Marten
>

Puzzled here. You mention that checksums won't match if LF is
not converted 'back' to CRLF.

But if the message and headers originally had (some) LF-only,
and/or had not already been modified to convert CRLF to LF, then
the checksum should match the message as-is should it not?

- Presuming the checksum had been generated against an
un-altered sample...

Bill