Re: object naming conventions

Nuno Souto wrote:

> It's *never* been helpful for anything. I never understood this
> "naming standard" thing. It reeks of the earlier DBMS systems
> where this sort of stuff was needed. With modern RDBMS systems,
> this hasn't been needed since the early 80s.

Back in the 80's we did not even had a data dictionary. Special utilities had to be run (with complex config files) to define files (aka tables) and columns. For each column you also had to create a 3 char code (unique within that file).

Dealing with files as oppose to tables, made for interesting physical design implementations (you can only have a single record struct definition per file).

Then you coded the following command block for something like a SQL select:
XXXA71EAA1000EAA2000... Where XXX is db id (or something) you could ignore, A71 was a code + subcodes to say you want to select data and from what table, E meant projection and AA1 was the column you're projecting and 000 meant without any condition (I recall that a 501 was an equal sign..). The actual values (literals) were send via a pointer to a data block.

Anyone else here worked with databaseses like "relational databases" Adabas and Sesam? :-)

That of course was not half of it. Imaging the fun of having to write re-entrant Cobol code (akin to writing DLL's/DSO's in Windows/Unix) and coding command & data blocks...

Yeah.. won't call it the good old days, but we use to bitch a lot less than some DB developers and administrators today.. ;-)

--
Billy