Interactions between conversions and character set conversions
Character set conversions come into play at two points during message enqueue processing: during header processing, and during message body processing. (By header processing here we mean any actual conversions of the character set in the message header, e.g., character set translation of personal names in addresses, not the MIME relabelling or other sorts of header line transformations that can also be configured via a
CHARSET-CONVERSION mapping table.)
After the message header processing character set conversion check and before any message body character set conversion check, the MTA consults the
CONVERSIONS mapping table (if present). If a message is to be routed to the conversion channel due to a match in the
CONVERSIONS mapping table, then normally character set conversion for the message body during this enqueue is disabled; it is expected instead that any relevant character set conversion of the message body will be incorporated into the conversion channel's conversion script processing of the message body parts. (The
imsimta test -translation utility may come in useful in conversion scripts for such purposes.) Alternatively, new in MS 6.3, the "
Preprocess" keyword may be used in channel routing entries in the
CONVERSIONS mapping table to specify that character set conversion of the message body (including
RELABEL operations) should be performed during the enqueue to the conversion channel (or alternate conversion channel) rather than being disabled as normal at that stage. Or yet another alternative: character set conversion of the message body for messages coming from the conversion channel may be configured, to convert the character set in the message body after the conversion channel has performed its operations; for such configuration, note that the
CHARSET-CONVERSION probe will not normally show the conversion channel as the source channel, but rather show the "original" source channel as the source channel.