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

[Nate Duehr]

Very interesting discussion, all.  I'm still catching up on all
of it.

I must admit this is a bit of an "escapist" exercise for me to
discuss this, since another 20 years of supporting continually
"broken" stuff is starting to look very "not interesting" in many
ways.  Some days, I'd rather stick hot twigs in my eyes.  Other
days, you just laugh and move on.  "This too, will happen again."

(Or better, Warren Buffet's quote, "What we learn from History is
that we don't learn from History."  So true, so true.)
The really NEAT bugs only come out about once a year.  You know,
the ones where the assumptions all collide into a really
strange/difficult to reproduce bug... those are fun to
troubleshoot.
But the overall level of what I dubbed "mediocrity" gets old
in-between the interesting ones.  Silly stuff like, "Your
application doesn't install correctly for multiple users on a
Windows platform."

And you sit here shaking your head...

Who'd write an installer (maybe full-stop there, since there are
good pay-to-play Windows installers that handle this type of
thing), you think to yourself as you look at the customer's
screen shots...  that doesn't know how to install properly for
multiple users after that's been a feature on Windows machines
for well over a decade?


You marvel that such a silly mistake happened, forward on the
ticket as a "feature request", and try to decide if you like that
stupid mistakes like that give you job security... or if you just
want to throw your headset at your monitor for how dumb it really
is.

Then you wait three weeks for a "Product Manager" to meet with
the customer to explain how "difficult" it is to handle
multiple-user installs on Windows, thinking... "Gee, that's
funny... a whole lot of applications do that just fine..." and
they promise it for a future release...

Yes... real example.  I won't name the product just to be "fair"
to my employer... but it's not like that same type of story
doesn't happen at EVERY company.  It's just silly, annoying,
frustrating, and downright boring (after you've seen similar
things happen multiple times over two decades, and the excuses
are always the same...)... and you start to question your sanity.


In any formal computing education, we're all taught that
computers are (of course) logical little boxes that only do what
they're told.  And of course, that means that any bugs can
eventually be wrung out of the system.  But what they don't tell
you is the rest of this "stuff" we're all discussing.  Re-writes
without code re-use.  Lack of standards so simple that things
like multi-user installs on a multi-user OS platform, are just
completely missed. Engineers assuming "the network" never has
errors (God this one bites me right square in the ass at least
twice a year)...


I guess I wanted to poke the ideas out there again to stir up
discussion because the longer I do this, the more the phrase
(paraphrased), "Doing the same things over again = the definition
of insanity," creeps up into my consciousness.

Don't get me wrong, I like this stuff. But the longer I do it,
the longer the list of "been there, done that" stuff becomes.
You see the problem starting and you can immediately give a VERY
CLOSE guess of how many MONTHS it'll take to correct something
that should have been 2 hours of coding time, up-front.  But
companies have almost zero "cultural memory" about past mistakes,
and hate admitting them in the first place... so you just chuckle
and know it's just Deja Fu.

(Deja Fu: Somewhere, somehow, I've been kicked in the head like
this before.)

Not trying to rant or judge "Software Engineers" too harshly.
Just hunting for a methodology that somehow captures this "We've
made this stupid mistake before" information and actually applies
it to the next 12 products... and starting to think it will never
happen at ANY organization... except maybe those with those "hard
goals" like Aerospace, etc.

(By the way, Aerospace isn't exactly not prone to such silliness.
 Garmin just followed up their wildly popular and well-designed
G1000 integrated cockpit system, with an announcement of the
G3000 cockpit system for light business jets yesterday... with
touch screens.  Hello... Garmin Engineers?  Any of your engineers
ever tried using their iPhone on a bumpy road?  Yeah...
turbulence & aircraft... you know... like during instrument
approaches in bad weather, right when you're reaching for the
touchscreen with nothing to "hang your fingers on" or any tactile
way to find a button or knob while NOT looking at the device so
you can FLY THE PLANE?  Bad, bad, bad .... sucky human factors
engineering on that one... prompting a "Really?!" from me when I
read it.  The G1000 uses tactile... "software driven" REAL
buttons... on the EDGE of the screen, to move around that
system...)

--
  Nate Duehr
  nate at natetech.com
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.lug.boulder.co.us/pipermail/lug/attachments/20091020/f261681e/attachment.html>