Hi all,
I've just committed the is_header functionality. The committed patch will use
the is_header field in the messageblks table only during insertion. Since we
have to make sure all headerblks are marked as such before we can use this for
retrieval, some transitional phase is required.
Perhaps we can fix dbmail-util to set all is_header flags on the correct rows,
or we can provide a simple sql/bash/awk script to do the same.
I've successfully used myself:
#> mysql --skip-column-names -B -e "select dbmail_messageblk_idnr from
messageblks group by physmessage_id" dbmail |awk 'BEGIN { printf("\nupdate
dbmail_messageblks set is_header=1 where messageblk_idnr in ("); } { if(NR %
200 == 0) { printf("\nupdate dbmail_messageblks set is_header=1 where
messageblk_idnr in ("); i=0; } else { printf("%s,",$1); }}'|sed 's/,$/);/' |
mysql dbmail
Which is quite fast, even for many, many messageblks.
This patch has been running on my production machines for several months now and
was part of the experimental debian packages.
--
________________________________________________________________
Paul Stevens [EMAIL PROTECTED]
NET FACILITIES GROUP GPG/PGP: 1024D/11F8CD31
The Netherlands_______________________________________www.nfg.nl