Message Headers and Content

The body of a message, that is, the message content, is built up from zero or more input files or procedures. The input files and procedures are read or invoked in the order specified in the item list passed to the mtaSend() routine. The message body is built up by appending the next input source to the end of the previous input source. A blank line will be inserted in the message as a separator between input sources if the MTA_BLANK item is requested in the item list. The MTA_MSG_FILE and MTA_MSG_PROC item codes are used to specify the name or address of input files or procedures.

An initial message header may be supplied from either an input file or procedure. The message header will then be modified as needed when the message is enqueued. The MTA_HDR_FILE and MTA_HDR_PROC items are used to specify the name or address of an input file or procedure. If an initial message header is to be supplied, it must appear in the item list before any MTA_MSG_FILE or MTA_MSG_PROC items. A blank line must be supplied at the end of the message header, or at the start of the first message-body input source. This blank line will automatically be supplied when the MTA_BLANK item code is specified in the item list.

The MTA_MODE_ and MTA_ENC_ items control the access mode and encodings applied to message body input sources. These items set the current access mode and encoding to be applied to all subsequent input sources that appear in the item list. The default access mode is MTA_MODE_TEXT, which uses text mode access. The default encoding is MTA_ENC_UNKNOWN, which results in no encoding of the data.

The binary access mode will not be applied to input procedures. The access mode and encoding item codes do not apply to input sources for an initial message header, which is always accessed using the default access mode and never encoded.

Input procedures use the following calling format:

ssize_t proc(const char **bufadr)

where const char **bufadr is the address of pointer to the starting memory location of the next piece of input data.

The return value is ssize_t, which gives the length of the input data. A value that is equal to or greater than zero (0) indicates success. A value of minus one (-1) indicates that there is no further data to return (EOF). Any other negative value indicates an error for which processing should be aborted.

The procedure will be repeatedly called until a negative value is returned, which indicates all input data has been retrieved or an error occurred.