On January 17, 2002 at 15:14, Eugene Eric Kim wrote: > What filesystems do not support the characters appearing in valid message > IDs?
Windows and VMS. Since message-ids are used for annotations, I get around this problem by encoding the message-id in quoted-printable format. However, for VMS, there are filename length limitations, which most message-ids easily exceed. I think I will not bother with trying to deal with filename length issues since the user base is very small for such systems. The quoted-printable trick should be sufficient for special character issues, but it does add a little overhead. > As a temporary measure, you could use a CGI program that took the message > ID as a parameter, and retrieved the appropriate filename using the > information from .mhonarc.db. That would resolve the problem of invalid > filenames by giving you a persistent method for retrieving messages > independent of the filename. This is inefficient, but will work. A more efficient approach is to use a db, or dbm, file that maps message-ids to pathnames. It is possible to hook such a thing in with the new callback functions added in v2.5. Perl has modules for interfacing with various db file types. --ewh
