The dupetrigger

MOTIVATION

Often, users commit their comments and articles multiple times if they
don't believe the comment has been inserted into the databases. This is
clearly one of the disadvantages of the otherwise clever caching strategy
used by Mir. Without proper handling, this duplicate content will enlarge
the database and make the appearance of the generated web pages be less
satisfactory. 


SOLUTIONS

There are two approaches to get rid of these duplicate entries
(hereafter called "dupes"). The first approach is the dupetrigger that 
ensures that no entries in a given table are the same. This is the approach
currently pursued in Mir. If this trigger cannot be compiled or doesn't work
for any reason, you may work-around this problem by issuing the following
commands on your psql prompt:

CREATE UNIQUE INDEX idx_content_dupe ON content (title, subtitle, creator, description, content_data);
CREATE UNIQUE INDEX idx_comment_dupe ON comment (creator, to_media, description, title);

Please note that this approach will work but the implications on the Mir 
software have not been tested, yet. Maybe Mir will respond with an error 
or with other unexpected behaviour, but it should do no harm to the
database. The dupetrigger has been tested with Mir and is currently being 
operated on de.indymedia.org.


HOW TO INSTALL THE DUPE TRIGGER

To install the trigger do the following:

# cp Makefile-dist Makefile

# edit Makefile
(and change pathnames and PGVERSION if neccessary).

# make
(This makes the trigger binary and places it in INSTALLDIR with permissions 644)

# edit postgresql.script
(And change the path to dupetrigger.so to INSTALLDIR)

# psql -Upostgres -f postgresql.script Mir 
(This installs the the trigger binary and creates a new trigger.)