Copy of my post:

Some discrepancies on the GDSG Units description (http://library.gnome.org/devel/gdp-style-guide/2.22/units.html.en):

1. "Kilobit" and "Megabytes"

In SI, written-out prefixes are always lowercase, so this should be "kilobit" and "megabyte". (http://physics.nist.gov/Pubs/SP811/sec06.html#6.2.2)  Only abbreviations are capitalized.  I'd change "Zero" to "zero", as well.

2. "The recommended units in the following table are based on the Systeme International d'Unites (SI) system of units."

Most of those units are not part of SI at all.  :)  (Only seconds, meters, and deg C).  I'd say something like "the recommended style for Gnome is the same as SI style rules"

3. "Kb"

If you're using k- = 1000 (which is usually the case for bits), the k is lowercase according to SI.  Also, I'd recommend not ever abbreviating "bit", since it's already short and "b" is ambiguous.  Writing out "bit" always is recommended by IEC and NIST as well (http://physics.nist.gov/cuu/Units/binary.html and http://www.iec.ch/zone/si/si_bytes.htm)  We should say "kB" for 1,000 bytes, not "KB" or "Kb", and "kbit/s" for 1,000 bit/s, not "Kbps".

4. binary multipliers K and M

This is a point of contention, but the standards say k- = 1,000 and M- = 1,000,000, no exceptions.  SI specifically says "These SI prefixes refer strictly to powers of 10. They should not be used to indicate powers of 2 (for example, one kilobit represents 1000 bits and not 1024 bits)."  This is also recommended in the Linux Programmer's Manual (man 7 units), used in the kernel, lots of FOSS software, etc.  As currently written, you define the multipliers in two conflicting ways on the same page (M = 10^6 and M = 2^20)  I'm sure this point needs discussion before changing, but I'd recommend following the standards and using Ki- = 1,024, Mi- = 1,048,576.

Most things should be measuring in 1,000s anyway (k/M/G), so it will still be familiar abbreviations, just with consistent, correct meanings.  The non-standard K = 1024 convention was not familiar to users anyway, and results in a lot of confusion.

As for making the HIG and GDSG consistent, is this already mentioned in the HIG somewhere, or is Bug #309850 *proposing* that it be mentioned?

Created attachment 132362 [details] [review]
diff of changes to units.xml

Here's a diff of some proposed changes.  I added some other things besides those mentioned in the email, too, like using Unicode °C instead of <superscript>o</superscript>C.  I'm not sure whether that's better or worse, considering things like DocBook-->man page conversion?

Not obvious from the .diff: I changed the "Kb" example to "kbit" and added a "MiB" example in addition to the "MB" example.

The wording starting at "The recommended units in the following table" could be improved.

I used "bit" as the abbreviated form of "binary digit", instead of the ambiguous "b".  Both are standardized, depending on who you ask (NIST/IEC like "bit" and IEEE likes "b" http://en.wikipedia.org/wiki/Bit#Abbreviation_and_symbol)

We might also want the written out form of "GiB" to be "gigabinary byte" instead of "gibibyte"?  The "kibi-" and "gibi-" prefixes seem to be the part of this standard that people hate the most.  http://physics.nist.gov/cuu/Units/binary.html  http://www.iec.ch/zone/si/si_bytes.htm

As I understand it, the style guide just describes how to write the units, not which units to use in certain situations?  So any discussion of "use GiB for memory and GB for disk sizes" would belong in the HIG instead?

The GDP style guide is outdated and unmaintained, so any changes are unlikely to be pushed there without a full rewrite.