I was involved in developing MARC.pm, the predecessor to the MARC::Record
module now in wide use.
In my prior employment, I wrote a PHP/PostgreSQL/Apache application that
served up a listing of our electronic journals, searchable by title
keyword, range of starting letters, and by a home-grown set of subjects
comprising roughly two dozen terms. It has been superseded at that
institution by a commercial service.
Most of the coding I have done has been for behind-the-scenes data
harvesting and munging--what I think of as utility programming. Much of
it resulted in single-use utilities that were quite specific to data and
context. I have pretty much always used open-source tools in my work.
In my present position (systems analyst, School of Health Information
Sciences at the University of Texas--Houston), I am using Python to do
automated analysis of full-text resources harvested from the web, with
an eye to assigning terms from a home-grown controlled vocabulary.
Sadly, I'm a Linux exile in a Windows world, so I'm running ActiveState
Python on XP, loading the results into MS SQL Server. Cygwin makes life
Roy Tennant made an important observation when he noted that librarians
should have some knowledge of programming if only to have a grasp of the
ease or difficulty of a particular programming task. We all know how
nice it is when the folks we program for have a clue about this matter!