OT, sickening goof up..

[Circlip]

Used to work for a company that installed, serviced and repaired expensive Danish Hi-Fi and Tv's. With the increasing use of remote functions on the equipment it became a nightmare trying to instruct a customer (Invariably the female half of the partnership due to the time of the day for installation) in the amount of time available, how to drive it.

One of the punch lines I always imparted before leaving was " Read the manual cos we'll be ringing in a couple of days and asking questions on it" After a suitable silence period, the ice was broken by a chuckle. This was then followed by the MOST important instruction, "IF you get tied up with all the command settings and it goes mammaries up, switch off and unplug the mains, count to ten, replug and start again, the equipment will have reset to manufacturers original settings. This would be the first thing I would do if called out on a service call and that would cost you £70"

Surprising how many could achieve remote programming via a quick telephone call.

Regards Ian.

[garagemark]

Don't ya just hate it when things like that happen? Our aluminum smelter starts all of the melt pots at '0'. There are 84 pots in each room, but it sounds like an odd number. Since everything electrically and instrumentation wise is balanced, some guys get confused as to where to go to do their particular work. Real PIA once in a while if it's a new guy or a tired hand, though I am on site and don't have to travel far to clear up the confusion. But still...

[Evan]

It's a consequence of using computers and lazy programmers. In the world of computers zero is a valid number. When counting and assigning quantities in a computer it is normal practice to use "arrays" of numbers. These arrays start at zero by default. In fact, in computer programming wonderland it became necessary to differentiate between the value of zero and the value nothing. So a new designation was invented to describe a value holder that contains no value. It is NAN which stands for "Not A Number".

[Arthur.Marks]

46 hours of travel to do a job that took 30 seconds then the same to get home again, but I did get to see a lot of movies that week.

How much does a small digital camera cost these days?? More than the alphabet can make it 'round the world on the internet!

[bruto]

How much does a small digital camera cost these days?? More than the alphabet can make it 'round the world on the internet!

That wouldn't help much if the picture was of the plug going into # 1 instead of #0, though.

[ckelloug]

Hi Evan,

Your explanation above is a mischaracterization. The definition of NaN comes from the standardization of floating point number formats and calculation errors in IEEE 754. NaN is returned when certain calculations overflow such as division by by zero. It was never associated with array indexing and not developed to differentiate between zero and no-data.

Normally a sentinel value is designated in the context of the problem for describing a no-data condition. NaN, NIL , and NULL are all used occasionally as sentinels for non-existent data but of the three, it's an "off-label" use for NaN because it can conflict with the detection and handling of floating point error conditions.

Additionally, arrays that are indexed from zero are only common in languages derived from C. Fortran and its ilk use 1 based array indexes.

Both zero and one based indexes simplify certain problems and the language runtimes deal with the consequences of the choice made by the programming language's designers. It's usually wise not to show zero based indexes to end users even if the program uses them internally since zero based indexes tend to confuse users.

While you've got the history and derivation wrong for NaN, I do agree with you that you can count on programmers to be lazy. . .

[lynnl]

....
... in computer programming wonderland it became necessary to differentiate between the value of zero and the value nothing. So a new designation was invented to describe a value holder that contains no value. It is NAN which stands for "Not A Number".

Never heard that one before. I've only heard of "null".

Or in the COBOL world, one can use either "low value" or "high value", depending on how you want things to sort out.

[tlfamm]

It's a consequence of using computers and lazy programmers. In the world of computers zero is a valid number. When counting and assigning quantities in a computer it is normal practice to use "arrays" of numbers. These arrays start at zero by default. In fact, in computer programming wonderland it became necessary to differentiate between the value of zero and the value nothing. So a new designation was invented to describe a value holder that contains no value. It is NAN which stands for "Not A Number".

Termed, as you know, "zero-based counting" - and a frequent source of programming errors as the mind slips in and out of zero-based versus one-based counting.

A colleague once opined, "All good programmers are off by one" - a wry comment on the phenomenon.

[Tinkerer]

That wouldn't help much if the picture was of the plug going into # 1 instead of #0, though.

Maybe maybe not and over all view sent may of shown 0 was the first socket and the corrective action could of been relaid. Also a few pictures of all the connections PRIOR to unhooking could of been used as comparators when re hooking all.

[Evan]

Your explanation above is a mischaracterization. The definition of NaN comes from the standardization of floating point number formats and calculation errors in IEEE 754. NaN is returned when certain calculations overflow such as division by by zero. It was never associated with array indexing and not developed to differentiate between zero and no-data.

I didn't say it had anything to do with array indexing. However, it most certainly does have to do with distinguishing between zero, null and an undefined numeric value. NaN stands for the latter and is not associated with calculation errors. Note that Nan does not represent infinity, either positive or negative but is truly undefined. It also does not represent overflow or underflow since those may be assigned values of infinity or zero or may raise exceptions that are not related to Nan. NaN cannot be compared even to itself since it is a non value for a numeric value data type. It is use to indicate an undefined value.

Normally a sentinel value is designated in the context of the problem for describing a no-data condition. NaN, NIL , and NULL are all used occasionally as sentinels for non-existent data but of the three, it's an "off-label" use for NaN because it can conflict with the detection and handling of floating point error conditions.

NaN as a sentinel value differs from the NaN as an undefined value in how the value is represented in binary. The two have only the name in common. The default when assigning NaN is to set the fraction bit non zero which designates NaN as an undefined value, not a signaling value.

Additionally, arrays that are indexed from zero are only common in languages derived from C. Fortran and its ilk use 1 based array indexes.

C wasn't introduced until 1972. BASIC used zero based array indexes when it was first introduced in 1964 and LISP used them in around 1958. Perl, Ruby and Java also use zero based array indexing. In fact, the majority of computer languages use zero based indexing, not just for arrays but for indexing in general such as character positions in string handling.

Both zero and one based indexes simplify certain problems and the language runtimes deal with the consequences of the choice made by the programming language's designers. It's usually wise not to show zero based indexes to end users even if the program uses them internally since zero based indexes tend to confuse users.

Agreed.