[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: Delegating ownership of namespaces named through schemers.org
On Tue, 31 Jan 2006, Andrew Wilcox wrote:
>Alex Shinn writes:
>> For a centralized naming authority, we'd setup some sort of
>> registration system at a server like schemers.org, and delegate module
>> namespaces to separate "entities," either organizations or individual
>> authors, presumably identified uniquely by a public key. This is
>> complicated, rigid and doesn't scale well.
>OK, let's not do that. :-)
right... I do think, however, that there need to be a couple
different levels of library. R6RS is (if I understand correctly)
going to delegate some functions to specific, named libraries
(libraries in the sense of bindings not visible unless the library is
specifically declared to be in use). And these are functions that are
required by the standard.
So I think that there really must be a "standard" namespace,
controlled by a central authority (the RNRS committee).
I do not think that the SRFI's deserve the same consideration, and
"auxiliary libraries" including the SRFI's should be named
I agree with the idea from PLT's libs that the library name should be
independent of its position in a taxonomical hierarchy. If I write
(Dillinger Vector-lib) and someone else writes (Petrovsky Vector-lib)
it should be just fine to find one in the graphics/3d area of the
taxonomy and one in the data-structures/realtime area of the taxonomy.
And should it be discovered over time by, say, digital signal
processing application builders that one of them is an excellent fit
with tools they're using, it shouldn't break stuff to add an alias to
it in the taxonomy under "DSP/data-management."
So, really, once we are out of standard and into auxiliary libraries,
we have two namespaces to manage; A canonical namespace that linking
and so on uses, where the names of the libraries are (Unique-author-id
library-name optional-version) or something like that, and a
taxonomical namespace where the names are for information purposes and
to help people find the library they need.
The former can be very very flat. In fact I don't believe it would
ever really need layers at all beyond the unique author identifier.
It should probably be digitally-signed code so that nobody can pretend
to be somebody else and spoof the libraries.
The latter only needs to be about as "organized" as usenet newsgroups.
So if I've written my vector library (Dillinger Vector-Lib 1.2) I
should be able to just add declaration lines to it indicating the
places in the taxonomy that I want to "post" it to so people can find
it, sign it, and publish it.