Prev: About NXP TDA8007 (smart-card chip controller)
Next: GNUARM multiple definition of AT91F_AIC_ConfigureIt
From: Eric Schwartz on 27 Sep 2007 12:22
Joerg <notthisjoergsch(a)removethispacbell.net> writes:
> AFAIU you need to have a VB copy on each PC it gets installed
> on. Sure, you can then remove it but that scrapes the fringes of the
> legit range. Also, IT folks dislike installing programs that don't
> come from major SW design houses.

You know, I'm not the world's most fervent Microsoft admirer by a long
shot, but I would find it hard to describe them as anything other than
a "major SW design house".

-=Eric

From: Joerg on 27 Sep 2007 12:29
Eric Schwartz wrote:

> Joerg <notthisjoergsch(a)removethispacbell.net> writes:
>
>>AFAIU you need to have a VB copy on each PC it gets installed
>>on. Sure, you can then remove it but that scrapes the fringes of the
>>legit range. Also, IT folks dislike installing programs that don't
>>come from major SW design houses.
>
>
> You know, I'm not the world's most fervent Microsoft admirer by a long
> shot, but I would find it hard to describe them as anything other than
> a "major SW design house".
>

It wasn't about them. It was about me (or rather someone more versed in
programming) writing VB Code. That would be a not so major SW design
house ;-)

--
Regards, Joerg

http://www.analogconsultants.com
From: Clifford Heath on 27 Sep 2007 21:46
Joerg wrote:
>>> Your are probably right. However, that's a "version control" scheme
>>> that is totally foreign to me.
> Yes. What were they thinking? In order to remain compatible it seems you
> must install the old versions along with the latest. Man, the FDA would
> rip me apart in the air if I ever designed something like that.

Let me guess - you design devices which have one, or maybe two, interfaces,
being mostly in simple master-slave relationships?

Now try considering a system of 30,000-50,000 separate components, each one
with a dependency on from 5 to 50 of the others, each one potentially with
flaws and outstanding change requests, and you get some concept of the
configuration complexity. We're talking hundreds of millions of times more
configuration risk (in terms of ramifications of a change in any one of those
interfaces) than you have in your devices.

I'm not saying your devices aren't more complex internally, or that each
interface isn't complex in itself, and it's probable that *that* complexity
is higher than each of the elements of a generalized software runtime, but
they still aren't even on the same planet when you consider configuration
risk.

It doesn't help that the Windows APIs are at least an order of magnitude
more numerous than strictly required... but that's exactly why the .NET
runtime was necessary... it forms a new platform that's vastly simpler
than the old one, and can move off Intel architectures to boot (vis Mono).

Clifford Heath.
From: Joerg on 28 Sep 2007 10:38
Clifford Heath wrote:

> Joerg wrote:
>
>>>> Your are probably right. However, that's a "version control" scheme
>>>> that is totally foreign to me.
>>
>> Yes. What were they thinking? In order to remain compatible it seems
>> you must install the old versions along with the latest. Man, the FDA
>> would rip me apart in the air if I ever designed something like that.
>
>
> Let me guess - you design devices which have one, or maybe two, interfaces,
> being mostly in simple master-slave relationships?
>

They are a bit more complicated than that but not five-digit. The
largest ones I design typically still have less than 100 locations that
need to exchange data. But some of those are realtime to the point
where, for example, another section of the whole system needs to trigger
on an ECG signal within spec'd accuracy. Failure to do so in timely
fashion can cause some grief.


> Now try considering a system of 30,000-50,000 separate components, each one
> with a dependency on from 5 to 50 of the others, each one potentially with
> flaws and outstanding change requests, and you get some concept of the
> configuration complexity. We're talking hundreds of millions of times more
> configuration risk (in terms of ramifications of a change in any one of
> those
> interfaces) than you have in your devices.
>
> I'm not saying your devices aren't more complex internally, or that each
> interface isn't complex in itself, and it's probable that *that* complexity
> is higher than each of the elements of a generalized software runtime, but
> they still aren't even on the same planet when you consider configuration
> risk.
>
> It doesn't help that the Windows APIs are at least an order of magnitude
> more numerous than strictly required... but that's exactly why the .NET
> runtime was necessary... it forms a new platform that's vastly simpler
> than the old one, and can move off Intel architectures to boot (vis Mono).
>

I don't dispute the need for something like .NET. It's just that I find
the implementation not to be as organized as I'd like it or as agencies
may request that our systems live up to.

--
Regards, Joerg

http://www.analogconsultants.com
First  |  Prev  | 
Pages: 1 2 3 4 5 6 7
Prev: About NXP TDA8007 (smart-card chip controller)
Next: GNUARM multiple definition of AT91F_AIC_ConfigureIt