Re: [Lwip] WGLC for lwig-terminology

[Carsten Bormann <[e-mail]>]

Hi Dominique,

thanks for the review and the good suggestions.

> - Section 2: "economically" -> "economically"
> - Section 2.2.1: what about explicating the MSL acronym as done for DTN in the following paragraph. I assume TCP does not need explicating.


I have performed these editorial changes in the SVN.  I also added a reference to RFC 793 for those people who want to look up the definition of the MSL.

(See http://trac.tools.ietf.org/wg/lwig/trac/changeset/11 for the complete set of changes I propose based on your comments.)

> - Section 3: Ki is not a standard prefix notation. K has traditionally been used for 1024 even though it is the SI unit for temperature. 

This used to be true, but is now outdated information.

IEC 80000-13:2008 (part of the series of standards documenting the SI system, which defines things like the meter and the prefixes like kilo and mega) defines "Prefixes for binary multiples", including Kibi, abbreviated Ki, and the Byte (8 bits, abbreviated o or B) *).

In the last decade, this terminology has found a home in most operating systems, and it behooves a terminology document to use the current terminology where it already has been accepted into the mainstream.

Of course, the reader may not be familiar with something that became the standard way of naming things only 5 years ago, so I also added a short explanation to the caption of the table.

> - Section 2.1, regarding the multiple facets: "- constraints on the maximum code size (ROM/Flash);", ..., "- constraints on the available electrical power". Do we want to mention CPU computational power as a possible constraint? RAM size is usually a limit before CPU power nowaday, but who knows.


CPU "power" may be a proxy for 
1) the amount of complexity a CPU can handle, or
2) for its speed.

1 is pretty much limited by its addressable RAM/ROM, which is covered.

Re 2: We haven't tried to provide terminology for CPU speed, and I doubt that we could come up with something very useful here.  The set of facets listed is intended as a lead-in to the next sections (I notice that we failed to provide a pointer to section 4 here, which I also added).
I also think that CPU speed is rarely by itself a limiting factor creating a "constrained device".

So I'm not as sure that we should add CPU speed to this list.  What do other reviewers think?

Grüße, Carsten

*) B is well-known, but also the SI abbreviation for Bel, and it also violates one rule for SI unit naming (there never was a Mrs. or Mr. Byte, which is a prerequisite to upper-case unit abbreviations).  o would be better in this respect, but, while o is the usual abbreviation in France, it is virtually unheard of in English-speaking countries.  Bel also only ever occurs in dB, and decibytes never occur.  So B is likely to remain the mainstream abbreviation for byte, much as we'd like otherwise.