Was FORTRAN buggy 4319

It was more of not having the QA reporting to the same cost center manager. To settle disputes of even small detail took trip up many layers of the org chart. However, I've been told recently that DEC politics was miniscule compared to what goes on now in other places. If this is true, I don't see how anything productive gets done.

No. The QA group supposedly did, but it was worthless as I described. We didn't have a machine to do that kind of testing. What we did do is use the monitor we built this week. The -10's job mix was BLISS compiles, MACRO-10 buttemblies, lots of TECO editing, lots of system services (printers, networks, magtapes, etc.). Once in a while a PCOPY would go on but that was usually done at night when the system didn't have to swap out the disk-disk I-O job.

This was a stale job mix. We picked our field test sites that

I don't think Lynn had to deal with as many variations as we did. Most of IBM's software (sources) stayed under their control. We didn't have that with TOPS-10. Until the new BLISS edicts, we did not keep our knowledge to ourselves; we shipped most sources. The only thing I can remember being restricted was MACY11.

We used it and JMF stared at SYSDPY stats.

The difference between response time of SMP and AP (I'm talking about user perception) was similar to the difference between a KA and a Pentium.

I believe all of this came after the development was done. You cannot measure a target moving at c. Since one of customers planned to have a 500 job monitor, we did builds and we would run it for a little while just to make sure Dwayne had chance of getting the system built and booted. Everything else was an exercise for the customer. This had to be the case because nobody ran the same monitor.

Those came in as reported bugs. The marketing machine would run its benchmarks. There were very few, IIRC, real bugs found with that. It was our experience that the real testing of things listed had to come from the customers. We did try to do some test suites that emulated a field test site but we caught more problems simply building their monitors without their patches and custom device drivers.

Most of your list was dealt with when JMF and TW planned their coding schedules. They already knew which areas needed work. If performance was a software project, it wouldn't get funded easily. All of our hardware projects were the ones that dealt with perfomance improvements. Note that performance improvements back then included increasing capacity.

The areas we worked on was not picked by DEC. It was picked by customers.

BAH