It is even worse than you think. When I was developing the HP 5334B counter, the plan was to use the firmware from the 5334A counter (currently in production) and use the same legacy microprocessors. However, we wanted to replace the expensive HP made divide by 10 prescaler with an off the shelf divide by 64 (or some power of two, I don't remember). At that point in time, we had already, while still building the 5334A, reached the point described below. No source code, and no development system. Also, the whole thing had been shoehorned into 2k of memory, with little or no extra space.
Fortunately, the original software engineer still worked for the company and I talked him into tweaking the machine code to change the prescale modulus. Miraculously, he was able to pull it off, and the rest is history. Speaking of important customers, the military wanted some feature called "MATE" which meant something like Military Automatic Test Equipment. It involved translating commands from HP-IB. What they had to do to add this was add a board with its own microprocessor that did the translation and wire it between the HP-IB hardware and the measurement controller, because they could not modify the measurement controller for the reasons given below, and other reasons. Rick Karlquist N6RK Magnus Danielson wrote: > On 10/24/2010 05:21 AM, David I. Emery wrote: >> On Sat, Oct 23, 2010 at 02:44:51PM -0700, J. Forster wrote: >>> Very true, exncept it's more like 5-10 years. >> >> These days John is absolutely right... likely none of the >> developers, none of the equipment, perhaps not even the corporate >> shell of the division or department that designed the product and >> wrote the software survives. Probably the source code was thrown >> out with the old servers that were sold for scrap... or just carted off >> to be shredded with all the other paper and electronic records... >> >> Horror stories abound about organizations that need to make some >> minor patch or change to source code of a popular product for some >> important customer even just a few years after its release and nobody >> can find the right source code or the right build environment >> (compilers, libraries, OS etc and the hardware they ran on) or if they >> can be found it takes many many hours of expensive time and talent to >> reconstruct the right stuff to actually make a code image that matches >> what is shipping. >> >> > > Even if you have the source code, you need to keep the compiling > environment in a runable state. This may be problematic since things > like licensees may have gone out (nobody payed for them and you can't > get a new one), the machines setup for running it has dropped out of > service. Maybe some manual work was never automated so some of the magic > is in the hands of someone that doesn't work there anymore... or recall > it. > > Keeping the source is one thing, being able to actually use it is another. > > The RTOS manuals and support files may have been tossed for instance... > > With all this... I still want the source thought. Alongside with other > internal documentation. I have never seen any of that surfaced for > boatanchors of any kind. > > Cheers, > Magnus > > _______________________________________________ > time-nuts mailing list -- [email protected] > To unsubscribe, go to > https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts > and follow the instructions there. > > _______________________________________________ time-nuts mailing list -- [email protected] To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
