Re: Package naming conventions

Nikodemus Siivola <[email protected]> writes:

> This is essentially return to "vn-cclan package naming conventions and
> repository organization" thread from May.

I've replied to the head of this thread, because I think it went off
on a bit of a tangent with CL packages, whereas we're actually
discussing cclan/asdf system names

(with-digression-anyway "Although I think 'fully qualified' package
names are OK if your style of programming is to use-package the things
you depend on, they rapidly get unwieldy if you'd rather use foo:bar
to pick up symbols in other packages - unless you have something like
hierarchical package support or per-package nicknames, neither of
which are in standard CL.  I'd like cclan to support both styles of
symbol reference without imposing this burden on users or
implementors")

>  * Flat namespace. Packages are free to name themselves, but
>    moderation is applied so that canonical names aren't frivolously
>    grabbed.

>  * Packages have category tags and a description. An CMUCL-specific
>    XML parser might belong to categories Text, XML, Parser, CMUCL.

>    Packages that don't have any implementation tags are automatically
>    tagged "Portable".
>
>  * Provide a tool to search by categories, etc.

My personal opinion here is that I'd like to see cliki in the role of
"human interface to cclan", with a cliki page of the same name as the
cclan package, for each package in the archive.  This would have the
following effect on your criteria -

Namespace: flat, name yourself, first come first served.  The
disapproval of the community (or, just as likely, the fact that someone
else has got there first) ensures that 'canonical' names aren't
frivolously grabbed.

Packages can be tagged using topic markers.  Presently the topic
markers are mostly hierarchical, but there's no reason not to use
cross-cutting markers as well.  CLiki searches can be written (perhaps
using a dedicated client; the syntax takes some getting used to, and
the parser has a sullen approach to error reporting) to find
pages^Wpackages with particular markers set.  See the Entomotomy bug
searches for some ideas.

I'm open to debate on this.  There's a formality continuum in these
matters, and different people are comfortable in different ranges
along it.  I just want to make sure the ad-hoc view is represented.


-dan

-- 

   http://www.cliki.net/ - Link farm for free CL-on-Unix resources 


-------------------------------------------------------
This SF.net email is sponsored by: VM Ware
With VMware you can run multiple operating systems on a single machine.
WITHOUT REBOOTING! Mix Linux / Windows / Novell virtual machines
at the same time. Free trial click here:http://www.vmware.com/wl/offer/358/0