Anyone out there using Ada ?

Whoa, I remember it as well. I did play with it for some time on my first 6809 based design (the "grandma"... :-). I had two 8" floppies which were moaning often for > 1/2 hour while compiling... Too bad I have lost the disks with it, I could run it on the DPS emulation of my th9 system (the "grandpa"... :-).

I cannot take seriously ISR code written in any HLL today either. I know there is a lot of such a mess being done every day, and it "works" because the silicon nowadays is >10x (usually well > 100x) overkill for the current level of software which is produced, but this does not make it any better.

You will probably remember the BASICM for the 6809, it did have specifically designed extensions for IRQ, FIRQ etc. handling... I have played with it as well, even used those while playing, I never put it into a project though. But it still can run on a th9 emulation I have in a DPS window. The guy who wrote it was Herve Tireford, Motorola had published a lot of his works - I used these extensively as learning literture and tools.

Dimiter

------------------------------------------------------ Dimiter Popoff Transgalactic Instruments

------------------------------------------------------

Original message:

... snip ...

Chris tends to knock down anything except Misra. That is another accurate Brinch-Hansen quote that I have never seen before.

Yes.

In that era my systems were all written in ISO Pascal, with some extensions. My systems could run interpreted pcode or machine language, on multiple machines, including the 8080 and the HP3000. In 1980 I was about to release a complete CP/M system when the PC hit the world. The system got pulled back, because practical use on the 8086 required at least 32 bit pointers, and I didn't have the time (or excuse) to develop that at the time. I lost everything about 10 years ago in a combination of disk failure and someone stealing my floppy backups. The remnants are posted in the download/cpm section on my page.

... snip ...

That was a poor quality system. Mine was re-entrant practically everywhere, and failures were obviously necessary and documented.

... snip ...

You probably can. Ada is available as part of the GCC collection, so now if you can use C or C++ or Fortran or Objective C via gcc, you can also use Ada.

I have the same experience. And I still use an in-house developed Pascal compiler for embedded programming. (It knows Modula-2 as well.)

Misra and other such efforts are certainly important and useful, but they're still akin to trying to make a loaded revolver safe to use as a hammer. I've spent most of my professional life writing C and I certainly don't hate it (I like it, for tye typical reasons people like it), but I cannot fathom why somebody would choose it for safety- related projects when there was a better choice available (and I don't mean C++ or Java!). As Brinch-Hansen says, it is not rational. And when the evidence shows 100 times more residual errors in DO-178B Level A code written in C than in code written in an Ada-based language, it is also not responsible.

Mike

That's not true at all.

... snip ...

Well, that's the impression I get. I may be mistaken.

^^^^^^^^^^^^^^^^^^^^^^^^^^^^^

Key words.

The last Pascal program I wrote professionally was in 1984. Not for an embedded system (we were using a very expensive J73B compiler targeting the 1750A for that -- but I digress), but rather a tool to support data analysis on a VAX. I needed a way to mask off fields of bits within an unsigned integer. No shifts. No AND. Very ugly code.

The bitwise logical operators were _the_ primary reason I switched to C for most of my coding almost immediately thereafter.

Today, I use C because it is ubiquitous. I don't know of a modern CPU whose ALU is at least 8 bits wide that isn't supported by a C (or NQC) compiler. Data analysis tools tend to be written in Python these days, though...

Regards,

-=Dave

You often are.

Well, unlike some around here, I try to avoid snittiness.

The qualities of the programming language itself is only one of the many criteria when selecting a programming language. The availability/price/quality of tools and the availability of developers with sufficient knowledge/experience with that programming language are often more important considerations.

Not rational from a technical sense, yes. However, CEOs and their upper-level managers have a different thought space: economic. They see programmer hours and cost per hour as huge expenses and seek ways to reduce one or the other. A large automotive supplier ( _still_ in bankruptcy after over two years!) saw India and its' abundant supply of cheap C programmers as a way to cut costs. So, the decision was made to drop Modula-GM, with a number of technical advantages over C, and program in C. To paraphrase Farragut: Damn the technical people! Full speed ahead!

~Dave T~

to

Since we're talking about safety related software in this subthread, maybe we should add sufficient knowledge/experience with safety related design and coding as well. And that's where not all programmers are equal. Sure, you can find gazillions of C and C++ programmers, but how much training will they require before they can perform adequately in the safety related arena?

It comes down to the oft-asked question, is it better to find programmers who know the problem space and development methodologies and get them up to speed on the tools, or programmers who know the tools and get them up to speed on the problem space and development methodologies? Assuming, of course, that you aren't able to find both in the same person. I tend to think that "I don't know language X" is a poor reason not to hire an otherwise qualified programmer, and "I know language X" is a poor reason to hire an otherwise unqualified programmer.

Mike

Full

Yep, I can believe that. Odd that it didn't occur to anybody to teach some of them Modula-GM. I doubt Indian programmers have a genetic propensity towards C and away from anything else.

Mike

Ahh, still thinking technical rather than economic--got to break you of that! It costs money to train them and they already knew C. And training would have been an open-ended proposition at best, undesirable and unattractive at worst (from their viewpoint). Their employment model (late 90's and maybe still) was to work for a few months then go to another company which would offer them more money for their experience, so training would have been open-ended. And learning MGM would not have helped them at their next company, thus undesirable.

We did use them for 3-4 efforts involving MGM, and never got even one of the previous people we had worked with even a couple months before. So we trained them every time.

~Dave T~

ey

to

=A0Full

OK, I'm going to try and think economic now.

Give them more money to stick around and stop the revolving door?

How'd I do? :)

Hey, I realize it's a messy picture. But what other branch of engineering lets the engineers use less efficient, more error-prone tools because they can get cheaper (up front, at least) employees that way? That's all part of the non-rationality pervading the whole question.

Mike