On 15/09/2020 11:31, The Natural Philosopher wrote:
> On 15/09/2020 11:15, Richard Kettlewell wrote:
>> The Natural Philosopher  writes:
>>> On 14/09/2020 14:01, Richard Kettlewell wrote:
>>>> Software is full of bugs.
>>>
>>> Yours may be.
>>> When writing embedded code you make sure it isn't.
>>
>> If you have a way to guarantee zero bugs in any nontrivial software, the
>> industry will beat a path to your door. Given you apparently can’t even
>> recognize a SQL injection vulnerability, I don’t think there’s much
>> chance of that happening any time soon.
>>
>>>> In a safety- and/or security-critical context,
>>>> you can’t just ignore them, you need to find them (preferrably before
>>>> someone else does), then recertify and redeploy the fixed software.
>>>>
>>>> Certification is a moving target, at least in my industry, I don’t know
>>>> about automotive. Same issues as above.
>>>>
>>>> Security is a moving target; attacks keep getting better. Same issues
>>>> again.
>>>>
>>>> And that’s before getting into changing customer needs, competitive
>>>> challenges, rebranding, ...
>>>
>>> In a car window winder?
>>
>> Bit more electronics in a car than just a window winder. At any rate
>> I’ll be trusting druck’s understanding of the costs and lifecycle or
>> automotive software over yours.
>>
> How much embedded programming have either of you done? I spent 5 years
> at it.
>
>
>
Look. My position is, and always has been, that *once the software has
been written*, the cost of deploying it is almost zero.

Neither you nor Druck have come up with more than in his case, proof by
assertion, and in your case, appeal to (his) authority, as to why this
is a false statement.

I am truly disappointed in you. I worked in manufacturing for a decade
before I stared writing firmware. I know all about product costs, and
amortising up front costs over productions runs, and I know whereof I speak.

Even in your far narrower experience, the upgrade cost of a new linux
application release is almost nothing. Apt-get update and apt-get
upgrade and then carry on and do something else. is all it takes.

Even a first time installations of linux is merely an hour compared with
the man millennia that have gone into writing it.

As far as upgrades go, how much is really necessary in a standalone
application? There are industrial applications running on Dos 2.2 out
there. The world runs on legacy COBOL and RPG that hasn't changed in
decades.

The mathematics of costings are very simple. Per unit cost is upfront
design cost divided by the production run, plus the opportunity cost of
implementing the hardware solution.

If as in the case of generic CPU enabled 'things' the production run is
in the millions, and its less than a man years worth of programming
effort, then the upfront cost is hardware  and implementation time plus
a millionth of a man year. Or about 5 man minutes or less per item
manufactured. (assuming 8 hour days, and 200 working days in a year)

It takes  a lot less than that to flash some ROM, EEAROM EEPROM  or what
ever the fashionable hardware is this year.
So, in such a case, the total cost of the firmware, installed is about 9
man minutes, or less. And even at the inflated rates that coders are
paid, that's not more than pence.

And to re-flash it is probably less than 4 man minutes for an 'upgrade'

Now ask druck to respond as to what in this reasoning is wrong, and how
I managed to survive in hardware design and software design for decades
if I was so wrong about costings?

I repeat, the point is that digital hardware is cheap and software *once
written* is even cheaper. Which is why we are all managing the world of
technology through less than ideal interfaces of touch screens connected
by wires rather than by pedals,  knobs and buttons connected by cables
and hydraulic pipes.




--
  “A leader is best When people barely know he exists. Of a good leader,
who talks little,When his work is done, his aim fulfilled,They will say,
“We did this ourselves.”

― Lao Tzu, Tao Te Ching

--- SoupGate-Win32 v1.05