[nSLUG] I am reluctant to file any more bugs until those already reported are fixed.

Michael David Crawford mdcrawford at gmail.com
Wed Jul 15 04:23:43 ADT 2015


I posted this just now to xcode-users at apple.com and
cocoa-dev at apple.com; at the end is my mail to Richard Stallman in
which I point out that Free Software suffers the same problems.

   *********

Please understand that I do not wish to get anyone in trouble.  It is
common for heads to roll over product defects.  That is not even
remotely my objective.

To be perfectly clear: this is a widespread, systemic problem in many
industries but in my own experience and that of many others it is
particularly bad in the computer industry.

The problem I see is that users pay for features not for quality.
That's what users believe anyway; it's not hard to convince them
otherwise but it's uncommon for high-tech products to obtain market
share because they are good, rather they succeed because they are
first to market.

I once consulted for Sony Ericsson Mobile Communications because they
once owned the smartphone market with Windows CE devices but then
Apple ate their lunch with the iPhone due to Sony Ericsson's poor
quality.

I have many gripes with Apple's product but this eMail is not the
place to discuss them.  However I request that Apple's engineers and
managers read every Radar bug I have ever filed -
mdcrawford at gmail.com, also from my amcc.com and atimi.com Apple IDs.

I was a Senior Engineer working in the role of "Debug Meister" for
Apple's Traditional OS Integration team in 1995 and 1996; despite my
love for my work I requested an internal transfer to PowerBooks
because I gave up all hope that our bugs would all be fixed.  I was
offered an internal transfer to Copland but declined it because I had
the sense that Copland would never ship.

While I was at first hired as a contract Script Monkey for MacTCP
1.0.1 in 1989, I was able to get sign-off to debug the test tool
because, as my Newfie ex-wife would say, "strm_echo was a piece of
work."

I went on to find a code generation bug in the MPW C compiler; while
regressing it I found that increasing the number of characters in
certain source code symbols led MPC C to crash.  To be clear: my
program didn't crash, MPW C crashed while building my source.

A couple years later I received a developer CD whose MPW release notes
clearly explained my bug then said "Don't do that".

By then I was with Working Software; it was on the ropes when I hired
on, and ultimately failed because my predecessor made a piece of work
out of QuickLetter, sued Working Software for failing to pay his
ransom but lost, stole the source code then shipped a competing
products with many of the exact same bugs as QuickLetter had at the
time he left the company.

In 1993 I wrote data analysis code for a particle physics experiment
at CERN.  I required seven weeks to come to grip with CERNLIB and its
associated tools such as PATCHY, sort of like a cross-platform JCL
only undead exhumed and reanimated.  I required three days to write my
own patch then four days to run my Monte Carlo simulation that
calculated the acceptance, or sensitivity of our detector.

Particle physicist require all manner of software but are hardly ever
to actually obtain any.  Every law of physics other than general
relativity - gravity - is to be found in CERNLIB yet despite decades
of wandering its Gordian Labyrinth I have yet to actually find any.

When my own patch was cooked and so ready to serve I asked my
collaboration's grad students and postdocs to have a look at my
source.  At the time I was still an undergraduate but had worked in
the software industry for six years.  They were all stunned at my
FORTRAN source's eloquence and beauty.

"The reason particle physics software is so hard," I growled angrily,
"is that YOU PHYSICISTS MAKE IT HARD!"

After Vietnam my father left the Navy, studied for his MSEE at the
University of Idaho in Moscow then worked as a Civil Service engineer
at Mare Island Naval Shipyard until he retired at thirty.

He wrote test plans.

Nothing but test plans, by hand on paper with a ball-point pen.

Here's what happens if you don't get your test plans right:

   The Sinking of the USS Gitarro
   http://catless.ncl.ac.uk/Risks/16.52.html#subj2.1

(Actually "Guitarro")

What I usually tell people is that "Someone sunk a submarine at the
height of the Vietnam War because he didn't read the instructions".
In reality, someone sunk a submarine because his test plan failed to
provide the instructions.

There is lots more I could say but I will write up a more compelling
argument than post it at:

    Solving the Software Problem
    a Taxonomy of Error
    http://www.warplife.com/jonathan-swift/books/software-problem/

However I will close with the mail I sent to Richard Stallman just now.

He is quite diligent with his email; in 1990 I had the idea he could
tell me how to get connected to the Internet.  His reply?  "If you
don't have a place to just plug in I really don't know."

   **********

Richard,

The ancient Greek play "Lysistrata" portrays some women who put a stop
to a war by refusing to make love to their men.  This has actually
been done several times, once recently in south america but I don't
recall the details.  I would be happy to produce the [needed citation]
but not just now as my dialup doesn't really work.

The problems I see with poor quality - not just in software but
hardware as well - are not specific to any one codebase.  It is
widespread in Free Software, Open Source and proprietary software.

For years I have worked diligently to advise others of ways they can
fix their code, as well as why they should do so.  My articles, essays
as well as mailing list and message board posts are quite popular but
also controversial in that there are many who do not welcome my
message of quality.

Quite a serious problem is that some of these faults are in systems
used by law enforcement.  For example some clever fellow escaped from
prison by sending an email to an employee at his prison.  Again I
don't recall all the details but would be happy to dig them up.

Similarly innocent people go to jail or even prison because of software faults.

My jocular outlook on life frequently leads to my own arrest.  Not the
last time I was in the slammer but the time just before that, I was
promptly ordered released on my own recognizance - that is, without
bail.  Even so I was detained for five more months before my case was
completely dismissed.  I think the world of the Clark County,
Washington Sheriff's Deputies but they were unwilling to release me
until they received my release order themselves, which somehow got
dropped on the floor.

No doubt you've seen Terry Gilliam's "Brazil".  That's happening in
real life this days and with increasing frequency and severity.

I could detail the problems I experience but there are so many.  It's
not just me; consider that Mozilla stored its email in a proprietary,
compressed database.  I lost my email database when my filesystem
filled up.  I did report the bug at http://bugzilla.mozilla.org/ but
when I did so I found many other reports as well as gnashing of teeth
because that bug had been reported years before but not fixed.

While the Mozilla developers - not just the Mozilla corporation but
its unpaid volunteers - are among the more-egregious offenders there
are many others.

Other problems are the introduction of "features" that I and others
regard as bug, and the deprecation of features that I and others
depend on for our livelihoods.

I am well aware of the Free Software community's opposition to Apple's
proprietary code, vendor lock-in and other evildoing but at least at
one time, Apple Computer produced very high-quality products.

But no more, and not for a long time.

I discuss this in:

   Apple's Deep Insight Into User Interface Design
   http://www.warplife.com/mdc/essays/jump-the-shark.html

The bug I found in 2012 was trivial to reproduce, however the steps to
reproduce it are quite obscure.  While I could help Apple fix just
that one bug by filing a report at http://bugreport.apple.com/ what I
really want is for Apple to clamp down on all the hookers and blow
that are readily available within Infinite Loop.

Again: these are widespread problems.  I find examples almost
everywhere I look.  Free Software is no exception.

However:

My satirical writing and my harsh, confrontive criticism in meatspace
aren't doing the job.  Recently I attempt gentle diplomacy.  To some
extent that seems to help, but I'm just one person.

Can you help in any way, or recommend something or someone else who can?

Ever Faithful,

Mike
--
Michael David Crawford, Consulting Software Engineer
mdcrawford at gmail.com
http://www.warplife.com/mdc/

    While Every Deity hath the Insight to Foretell the Future
    Even G-d Almighty Himself Possesseth Not the Power to Undo the Past.


More information about the nSLUG mailing list