Re: Some slides and a paper from L.C.A. 2004

[George <jirka 5z com>]

On Wed, Jan 21, 2004 at 12:27:36AM +0000, Alan Cox wrote:
> On Maw, 2004-01-20 at 15:31, Richard Stallman wrote:
> >     (1) The paper for the Linux and OSS in Government software:
> > 	    http://www.gnome.org/~malcolm/a11y/
> > 
> > In connection with GNOME, please remember to speak of it as
> > "free software", and to call the system "GNU/Linux".  Many people
> 
> With respect to the use of the term GNU/Linux please use Linux with the
> (TM) or the trademark symbol in situations where it might otherwise
> imply things like ownership (eg GNU/), otherwise it may lead to
> misleading assumptions about the Linux name

But that could lead to an ambiguity as in GNU/Linux(TM), which can mean two
things, GNU / Linux(TM) or GNU/Linux (TM).  Perhaps we need to add a set of
parenthesis as in: GNU/[Linux(TM)].  The notation is also confusing as the
system is really the natural homomorphism from GNU to GNU/[Linux(TM)] (then
[Linux(TM)] is of course the kernel of this homomorphism).  And so
GNU/[Linux(TM)] is rather what you get after you apply the whole operating
system to GNU.  To avoid this further ambiguity I think the entire system
should be called (using LaTeX as ascii is now deficient)

\phi : GNU \rightarrow GNU/[Linux(TM)]

When one wants to purely refer to Linux(TM) as the kernel, we can shorten
this by just \ker(\phi), which is 1 letter longer, but only in ascii.  When
printed it actually comes out as 6 characters as opposed to 9 in Linux(TM).

Now furthermore the above is abuse of notation since Linux(TM) is obviously
not part of GNU, but rather must be mapped into GNU and we can call this
map i : Linux(TM) \rightarrow GNU.  So really the system should be

\phi : GNU \rightarrow GNU/i(Linux(TM))

which is actually good since we can get rid of one set of parenthesis as
there is no ambiguity on where the (TM) applies.

Of course this all assumes that i(Linux(TM)) is normal in GNU which some
people would disagree with saying that only Hurd (which is itself a subgroup
of GNU) is normal in GNU.  But we can define i to map Linux(TM) onto Hurd,
then of course i(Linux(TM)) is normal in GNU and then of course \phi maps
Hurd to 0 since obviusly we're not using Hurd.

> (Sorry to bore everyone ****less but there are obligations to defend a
> trademark or lose it, therefore I have no choice but to remind Richard
> *again* and others)

But boring everyone is much more fun then being actually productive ...

George

-- 
George <jirka 5z com>
   I thoroughly disapprove of duels.  If a man should challenge
   me, I would take him kindly and forgivingly by the hand and
   lead him to a quiet place and kill him.
                       -- Mark Twain