This directory contains a package which documents an Rdb database in a format suitable for various PC and Macintosh word processing programs. I originally wrote a DCL command procedure which read the output of SQL and formatted it into a plain text file. However, I discovered that it was fairly easy to write the information in RTF format instead. RTF is the "Rich Text Format", developed by Microsoft, which can be read by many word processing programs on IBM PCs and clones, on the Macintosh, and probably other platforms. There is no license or royalty needed to write a document in RTF format. By writing in RTF format, the output of the procedure is a document which contains a header, page numbering, column headers, bolding and underlining, and so on: but the document is also revisable within the word processor. It also makes it possible for us to include the database documentation within our existing application documentation (which is already in a Macintosh word processor) without having to do a lot of re-formatting of plain ASCII text each time the database changes. The procedure was originally entirely in DCL (RTF_TO_RDB_DCL.COM is the last version in this form), but when I added the cross reference of domain usage I found that SQL did not give the original domain reference for views. I had in the past written some experimental programs (embedded SQL in C) which extracted information about a database directly from the database metadata. I adapted one of them to do the cross reference, and then moved some additional processing from DCL to the program. It is now about 50-50 between the program and DCL. WARNINGS The procedure is intended to work on our databases, where all fields (columns) in a relation (table) are based on domains (global fields). I don't know what it will do if it hits a table containing original field definitions. The list of tables will write up to 90 tables in two columns on one page. I don't know what will happen if the database has more than 90 columns. I originally wrote the new section information on the first page of each section manually. I decided that it would look a lot nicer if the name of the database and some other information appeared in a header on each page. I am still working on getting the header and footer (with the page numbers) to come out right; it works for some programs (WriteNow) and not for others (Microsoft Word). There may also be some oddities in how many blank lines there are at the top of each page. However, even with some hand touch-ups, it does nearly all of our documentation work automatically. It's been a big step forward for our documentation. Bart Z. Lederman System Resources Corp. Burlington, MA