|
All,
I was one of the balloters of 260.1, so I think I need to comment (since
I maintained a negative ballot on this standard). While I agree with Ashley that
this standard provides an excellent source of information on units (and I
provided a copy to John Ralph to assist in this area), I want to comment on the
use of the "-ibi-" prefixes (since Ashly mentioned them in his email). I am
firmly opposed to use kibi-, mebi, gibi, etc. as they are artificial
constructions to avoid "overloading" application of
prefixes.
Under 260.1 a kilobyte is exactly 1000 bytes rather than the 1024 (2^10)
bytes the industry is used to. If you want to specify 1024 bytes, you now have
to use kibibyte. Similarly, a megabyte is now 1,000,000 bytes rather than 2^20 =
1,048,576 bytes. Now, 2^20 bytes must be called a mebibyte. I think this is
nonsense and said so in my ballot. The only defense the BRC provided for the
change is that the terms were passed down to them from ANSI. Sorry, that's not
good enough. I am opposed because I do not believe the industry has any
interest in changing from the traditional metric prefixes and that the use of
the "old" prefixes is well understood and generally used. (How many of you say
you have 512 mebibytes on your PC? We can't say 512 meg any more. It would be
512 meb. How silly is that?) If there is any fear of confusion by users, we
simply specify in our standards that we are using the traditional prefixes and
define what those traditional prefixes mean.
John
-----
Dr. John W. Sheppard
-- Fellow, ARINC, 410-266-2099,
jsheppar@arinc.com
--
Assistant Research Professor, Johns Hopkins University, 410-516-4957,
jsheppa2@jhu.edu
All
I have just obtained the
updated version of IEEE 260.1 and believe this forms an excellent basis for
the list of permissible or acceptable units.
It defines the SI units
and SI derived units.
It lists many other deprecated units which I suggest
we avoid except where members of the WG can support their use. For example ppm
is deprecated but perhaps we should include it as it is in "common use", even
though we could use µmol/mol as I believe Joe suggested.
We should not
use unit symbols to mean more than one thing, for example Gi and Gb should not
be used for the unit Gilbert (deprecated) as these symbols are used for Gibi
(Gi) and gigabit (Gb).
This standard lists the "customary inch-pound"
units which many legacy program require.
The standard also lists the
acceptable alternative symbol for use with limited character sets. For example
Ohm is an acceptable alternative for the omega character, u for µ, and sec for
".
It provides a standardized method of indicating superscripts and
subscripts where the character set does not allow this, for example by using
the caret (^) and underscore(_).
It lists the binary prefixes (Ki, Mi,
Gi, Ti, Pi, Ei) which we should include.
We should avoid the use
of units which represent more than one quantity such as gal (gallon). Billion
and trillion also fall into this category. If it is necessary for us to
include gallon the appropriate suffix should be included (gal_US or gal_Imp).
The US and Imp are subscripts.
It also provides guidance on the use of
the bel (B) and neper (Np). Any reference level should be provided in
parentheses after the unit symbol. dB without a reference level may only be
used for relative powers.
I suggest that this standard answers most of
our questions. It may also discourage some common errors such as the
measurement of time in siemens (S) or as a plane angle (sec). I have spotted
three unavoidable dual uses so far; in the use of min, plane angle (where ' is
not available) and minute of time; in the use of C, degree Celsius (where ° is
not available) and coulomb; and in the use of F, degree Fahrenheit (where ° is
not available) and farad. Hopefully context should resolve this one.
As we are currently updating 1641 (we have a PAR now for P1641a)
we will take the opportunity to update our table of units. We would like to
have this aligned to the use in ATML.
One final point. The method of
describing the signal, or the description of the method of observing the
signal (e.g. pk-pk, average, ac, dc, rms, etc.) are important issues to be
covered. 260.1 allows them to be used after the unit separated by a space.
They have nothing to do with units themselves and should be kept
separate.
Ashley
|