[lug] can't make this stuff up, folks...

Nate Duehr nate at natetech.com
Mon Oct 19 10:20:23 MDT 2009


So my experience agrees with all of yours...

The newbie "engineer" codes up their "90% solution" for all the reasons
already presented.

This EVENTUALLY needs to be the same real 100% solution that was
available through a library, they just don't KNOW it yet... so
eventually they end up making a new "100% solution" over time.  Maybe in
a different programming language, because the new one is "the cool kid"
this week.

The dirty part: Each one of those "100% solutions" has serious bugs in
corner cases that haven't ever been dealt with yet... so they're not
really 100%.  They're a lot closer than the newbie's original
implementation, however.

Meanwhile, the 90% solution from the newbie gets touted by them, their
managers, and everyone involved as "THE NEXT BIG THING!" from the
software engineering group.  You know what my IMMEDIATE response to
brand new code not written by the grizzled veterans is, in a tech
support role, "Oh ****, here we go again."

Anyway, back to the abstract examples.  After lots of newbies do the
above, on a big picture level, we end up with thousands of things that
LOOK like "100% solutions" available, that if used by someone
experienced, blow up in their face... Tkil, would you say every library
in CPAN is bug-free and peer-reviewed CAREFULLY by many people?  I doubt
that... but then again, CPAN is a couple notches better than most
repositories...

Then, as tkil and others have said: Newbies never use them because
they're "too complex".  (They should be thinking, "Hmm... if it's this
complex, perhaps there's a reason.  Wonder what I'm missing here.") Or
worse, they don't know about them.

And the cycle repeats.  Over and over.

More and more "100%" solutions that really aren't -- are written every
day.  In different programming languages, even.  Our own "Tower of
Babel".

That about cover it?  :-)

--
  Nate Duehr
  nate at natetech.com



More information about the LUG mailing list