.Dd $Mdocdate$
.Dt mboxu 5
.Os
.
.Sh NAME
.Nm mboxu
.Nd readable mbox format
.
.
.Sh DESCIPTION
.Pp
.Nm
is an mbox-like format, in which messages are encoded in UTF-8.

.Nm
also forbids the inclusion of multipart messages inside the mbox database.
The body of a message must contain a single
.Em canonical
part in plain-text UTF-8.
Other parts, such as images, should be stored in separate local or remote files and referenced via a
.Ql Reference
header:
.
.Bd -literal -offset indent
Reference: att/review.pdf
.Ed
.
.Pp
Paths in the
.Ql Reference
header are relative to the location of the mbox database, unless they begin with a forward slash.
If a path begins with a protocol, such as http:// or ftp://, it should be regarded as remote.
.Pp
Messages are separated by a complete and correctly formatted
.Ql From\ 
line.
Lines in message bodies beginning with
.Ql From\ 
are not quoted.
.Pp
.Nm
is not particularly suitable for storing
.Dq standard
internet mail.
Rather, it should be used for more specific purposes, such as the storage of local data or of mail propagated in a specialized network not meant to be interoperable with the general e-mail network.
.Pp
Despite its differences from the standard mbox format,
.Nm
is readable by most modern mail readers.
.
.Sh RATIONALE
.Pp
RFC 4155 requires messages stored in mbox databases to follow the standard syntax and format of e-mail messages defined in RFC 2822 (but without the carriage return).
It also requires messages to be seven-bit text, such as ASCII or presumably UTF-7.
.Pp
This makes it difficult to use mbox for anything other than storing received mail on an internet mail server.
.Pp
.Nm
simplifies the mbox format in order to expand its usage to the storage of other data than real internet mail.
It has two goals:
.Bl -enum
.It
to be easily writable and readable by humans, and
.It
to be easily generated and parsed by machines.
.El
.
.Sh AUTHORS
.Pp
.Nm
is created by John Ankarström
.Aq Mt john (at) ankarstrom.se .