#12 CJAN Project Objectives - Indexing

[Michael Davey]

Obviously the packages will need to be indexed. It
seems obvious to build indexes based on every field in
a 'packing slip' file to be searchable as fast as
possible. The CPAN, for example is searchable by Title,
Author, & Functionality. Instead of names, perhaps
domains would be more appropriate when dealing with
packages.

[Robert (Jamie) Munro]

Logged In: YES
user_id=60940

Allow anyone to apply any category that the software is
relevant to and allow them to weight their rating. Then,
average everone's ratings. If someone is giving ratings
that disagree with the average by a lot, then reduce their
importance in determining ratings later.

[Nobody/Anonymous]

Logged In: NO

It seems legitamte for many authors to share a domain.
So I guess you need _author_.

However as code is "named" by domain, you also need domains.

RJMunro's voting idea has merit.

But I think there are two stages:
1) defining categories (perhaps use ACM categories ? :-)
2) applying categories to code/documents/tools/...

I suspect that the categories have two parts:
a) "material"
java source
other source
object code (class/zip/jar/jeff/...)
documents
external "references" or links
b) topic
class loader
bytecode verifier
disassembler
...

Note I think the second class should be a carefully designed hierarchy, allowing more specific categories to be added without invalidating earlier broad classifications.

- JonT

[Anonymous]

Logged In: YES
user_id=55979

SUN suggests that "Companies use their reversed Internet
domain name in their package names." So, would (or should)
authors share domains?

How about indexing "projects" this way:
a. by author
b. by topic (CPAN calls this category)
c. by domain (e.g., org.apache, org.cjan, etc.)
d. by project name (alphabetical)

Also, As Chris pointed out to me, CPAN modules can be
indexed by name without being assigned to a category. I
think this is a useful precedence to follow.

Im still not sure about the categories: its a top-
down/bottom-up thing for me. I can see the value of a well-
thought out hierarchy, but I also think that assigning
categories as they come in keeps the system leaner.

Could you point us to the ACM categories? Also, it might be
worth pinging the CPAN maintainers to see how they came up
with their categories. I original thought that a rating
system might make sense, but Im leaning against it after
discussing this with Chris.

Lately, do we need to categorize material or just index
it for searching capabilities?