Citylink - useless, useless, useless.

[John Rumm]

Yup, talking at (slightly) crossed purposes. This was a case of what was working VHDL failing to produce a working result on the new device. The binary was obviously recompiled for the new device.

That was a comment on the general state of embedded tools market rather than about that specific case (and also more focussed on software rather than programmable logic)

i.e. it is not uncommon to find them producing results at times, generating unexecutable code for example, or simply crashing with no audit trail at build time.

[djc]

on Trusting Trust Ken Thompson Reprinted from Communication of the ACM, Vol. 27, No. 8, August 1984, pp. 761-763. Copyright © 1984, Association for Computing Machinery, Inc. Also appears in ACM Turing Award Lectures: The First Twenty Years

1965-1985 Copyright © 1987 by the ACM press and Computers Under Attack: Intruders, Worms, and Viruses Copyright © 1990 by the ACM press.

[Bob Eager]

Yes, I have that issue of CACM somewhere. I was told about it rather earlier; it was mentioned at some conference.

[Windmill]

That sounds like instructions written in binary to a drum, then later executed in sequence as the drum rotated. Before my time, but there was a description in an addendum to the Jargon File (or maybe ESR's transcription of that into book form).

[Bob Eager]

Not at all. The instructions were loaded off the hard disk (a reserved area) and placed into very fast RAM, presumably static RAM. This machine would have been designed in the early 1970s...we had one of the first, and it was delivered in summer 1976.

[Andy Champ]

Huh. Do you think I made a mistake sending my son off to do CompSci at Uni? Bob will undoubtedly say it was exactly the right choice...

Andy

[John Rumm]

Update in case you are interested:

In fact, it turns out we both were at slightly crossed purposes. I checked earlier today, and it turns out it was actually the *same device* (a Xylinx Spartan), with the same functionality in the VHDL.

What had happened, was since there was plenty of spare capacity and IO available on the device, some extra functionality (completely unconnected to the working stuff) had been added, that worked ok, but for no apparent reason caused a half clock shift in some of the IO of the bit that used to work.

Needless to say, to investigate they brought some internal signals out to spare pins for debug purposes, and (as you probably guessed) that fixed it anyway.

[Bob Eager]

The subject coverage has certainly changed. There is new stuff all the time...so something has to give if courses are to keep up.

We don't do much on assembler, for example - but 25 years ago theer was much less on networking and nothing on web stuff.

Generally, most courses give a good basic coverage, adhering to:

rest of most courses is research driven. The real winner for students is practical experience - either a Year in Industry, or failing that, we offer something most places don't:

it may have been the right choice - and in your case, of course, the right Uni! .-)

[Huge]

The son of my best mate (I went to UKC, as did my best mate (we met there)

- although neither of us did Comp Sci, although we both distantly know Bob) went to UKC to study Comp Sci and thinks it's bloody brilliant. (*)

Well done Bob!

(* They have extraordinarily little timetabled stuff. How do they learn anything?)

[Bob Eager]

Curious to know who, but will understand if I'm not told!

They get about 8 hours of lectures a week, and up to about 6-8 hoirs of small group teaching. Add to that time for going over lectures, doing assessments, group meetings, reading, general stuff...it's about a 40 hour week. Staff are generally available on demand (open door policy) and of course we answer email and also anonymised questions.

'Learn by Doing'!

[Huge]

YHM.

[Bob Eager]

Ah yes. Of course.

[Man at B&Q]

LOL! I would be looking at the timing constraints or reset states.

MBQ