Guide to Image Class Documentation
See DLXS, Getting
Started for installation of DLXS, including sample data.
Read the Image Class workshop
tutorial for a useful introduction to working
with Image Class.
Image Class Search Engine Options
Image Class offers two options for metadata indexing and searching.
Option 1: MySQL is the
preferred choice because Image Class data fits well into the relational database
model. MySQL is free.
Option 2: XPAT is
an excellent full-text, SGML/XML aware search engine that works fine with Image Class. However, since MySQL works just as well and is easier to use (from a collection implementation perspective), XPat is not recommended and may not be supported in future versions of DLXS. There is a free version
of XPat, but larger databases may need the full version, which has a license
fee. Working with XPat has a steeper learning curve and the data preparation
process in general is more cumbersome than for MySQL. However if XPat is already
being used for other reasons, maybe it is the best choice for Image Class as
well. At Michigan, MySQL is used with Image Class.
Not To Be Missed
- MySQL, by default, indexes words 4 characters or longer. Shorter
words are therefore not searchable. This can be changed if desired. It
requires having the following configuration directive in the [mysqld] section of the MySQL configuration file (usually /etc/my.cnf):
Once this is done, existing fulltext indexes will need to be
rebuilt. The following command does the job nicely in MySQL.
REPAIR TABLE tbl_name QUICK
- There are other situations where you might need to repair tables or
rebuild indexes. Migration of tables from one version of MySQL to
another can be such a case. Signs of trouble may be subtle. Boolean
searching has been known to fail quietly. In addition to the REPAIR
command above, you can try...
REPAIR TABLE tbl_name USE_FRM
Thanks to Aaron Brenner for the tip and this link http://dev.mysql.com/doc/mysql/en/Fulltext_Fine-tuning.html.
- The amount of time it takes for thumbnail images to display in the browser can be reduced dramatically by creating a symlink for each collection like this....
ln -s $DLXSROOT/img/c/coll/index thumb
The only drawback to this approach is that it bypasses the cgi and authentication for thumbnails by essentially putting them under the web document root. If thumbnail security is important, don't create the symlink.
A Common Problem
- If you get a division by zero error from the Image Class CGI when trying to view a full size image, double check the installation of mrsid_retrieve. Most likeley, it is not installed properly.