🚀 go-pugleaf

I'm not sure I'd classify it as either a toy or a RDBMS
(where R stands for Real). Rather it's another tool which
can do reasonable work with minimum effort on files which
have a known number of fields.

The "data base" is a standard ASCII file, indices can be built
if desired. Joins, selections, projections are supported.
The DDL is a simple ASCII file in either the current directory
or the directory where the "data base" is stored - thereby
allowing a regular UNIX user access to files such as /etc/passwd
where they do not normally have write permission into /etc.
(Maybe you want to join the passwd file to something else)

We've found it useful for subsetting password-type files.
We also wanted to build/maintain some automatic cross-referencing
of IMS data, and the easiest way still (after 2 years no less)
seems to be to dump the data into UNIX and throw a dozen or so
tools at it. UNITY is the central tool in this instance.

UNITY also provides for field validation - after the fact. This has
proven useful because I often want to find out "what records aren't ...".
Again, an ASCII file defines the validation syntax.

Much of what UNITY does could be handled with awk/cut/sed/bs.
Then again, I suppose one could do the whole thing in C.
UNITY doesn't win any speed records, but neither does awk for that matter.
It *IS* moderately easy to master and use - and is appropriate when you
want to work on files which for one reason or another cannot permanently
be placed into one of the commercial DBMS'.
-- 
David St. Pierre		{ihnp4,dual,qantel}!ptsfa!dsp