[thelist] Re: higher-gen db's... sigh

Techwatcher techwatcher at accesswriters.com
Fri Apr 26 17:02:00 CDT 2002


Hi, all

Don't mind me, I'm just depressed at the end of a long, bad week...

> From: "Jay Blanchard" <jay.blanchard at niicommunications.com>
>
> I have a table where I want to find duplicate records, and eliminate
them
>
> tblFOO
> ID     Name     Category     LastDate
> 1      Tom      HR           2002-03-05
> 2      Dick     R&D          2002-03-05
> 3      Harry    Mgmt         2002-03-05
> 4      Tom      HR           2002-03-05
> 5      Moe      Mgmt         2002-03-06
> 6      Larry    BOD          2002-03-05
> 7      Curly    Support      2002-03-06
>
> SELECT Name, Category, Date
> FROM tblFOO
> WHERE Name = Name
> AND Category = Category
> AND LastDate = LastDate
>
> returns all records, because 'Name = Name' is true, as are the other
items.
> In the table where I am doing this I would not know any of the
values, so I
> cannot be more specific about criteria. The ID number for each record
is
> unique. This is in MySQL, so there is no UNION available. Has anybody
done
> this before?
>
> TIA!
>
> Jay

This is the kind of problem for which APL or J (among many others) has
literally one-line solutions. That's why I still keep them on whatever
computer I'm currently using: to solve this sort of problem.

I wonder whether dp/it/software engineering educators will eventually
figure out that canned db programs can't do the whole job. This rant is
also in reference to the fellow who had to use Mysql (or was it sql?)
to draw from two "legacy" databases.

Once I wrote an (online) documentation tool for Merrill Lynch traders
(as consultant, not employee). I used APL, and it took whatever text or
fields you wanted to give it, then allowed a free-form, English-
language search on the resulting db. That took about 12 weeks,
including a week or two of figuring out where I could find some soon-to-
be data (this ops techie had been keeping notes about new software
coming in), who the experts were (to interview), and the initial
relevant dimensions for the software tools I was documenting.

Took another week or so to work out my structure and design issues, and
another week at the end to teach an employee programmer about the
system, the tools I'd created to update the db at intervals (or ad hoc,
when anyone knew new tools had been added to the system).

Then the mgr said, "Oh, but we wanted it for SQL." Took another whole
week to write a few more programs to strip the descriptions down to
keywords, make sure the keywords weren't showing up (individually) in
most descriptions but were (in aggregate) hitting every description...
and present him with a much less functional SQL-compatible db. Also, of
course, to build new front-end, and design (and print on card stock)
the quick-ref card for users (keywords and how to input them).

I know we have to have dumbed-down word processors (for the
secretaries), but do we have to rely 100% on dumbed-down db languages
too? Even us professionals?

sigh. mope.



More information about the thelist mailing list