bsa@ncoast.UUCP (Brandon Allbery) (03/10/85)
I've been looking over the 2.10.2 news sources (we are about to change over,
finally), and wondering about new news ideas. Here's what I have come up with
so far; I'd like some input as to how easy/hard it would be (I will try it
after the news srcs are freed up), compatibility issues, etc.
Basically, the current newsgroup-image spool directory would be scrapped.
Messages would be stored in /usr/spool/news by their message-id (minus <>).
The history database (history.{dir,pag}) would take over those duties,
using message-ID as the key and a comma-separated list of "newsgroups" as
data. User .newsrc's would then become databases like the history database.
The subject line could also be stored as data, provided that dbm(3) (which
I do not really understand, since I'm a wizard-in-training on a binary site)
can take it.
Keyword-driven searches can then search for newsgroup and/or subject, or
searches by message-ID can be done, including all articles originating on
such-and-such a machine. Duplicate article rejection can be automatic,
since if its message-ID is in the user's .newsrc it can be ignored. News
would run faster and be easier to use.
I haven't spent a whole lot of time on this, since I have a large number
of other things to do, but it seems to me that this would make things easier
for the news reader and simplify the naming problem specified by Chuq as
well.
Newsgroups? Who cares? It becomes a sort of general subject, nothing more,
and such things as cmsg newgroup, etc. become obsolete, anbd users can make
them as they see fit; but the distributions part would still have meaning.
Thus only the "fa" in "fa.telecom" (for example) would be paid attention to
by the software. Of course, mailings should be encouraged to discourage the
misuse of such groups, just as is often done now.
This is not intended to solve all problems, it's just a start on simplifying
the whole thing. ANy comments, suggestions, flames, etc., send 'em here.
--bsa
--
Brandon Allbery, decvax!cwruecmp!ncoast!bsa, ncoast!bsa@case.csnet (etc.)
6504 Chestnut Road Independence, Ohio 44131 +1 216 524 1416 -- CIS 74106,1032
-=> Does the Doctor make house calls? <=-alb@brunix.UUCP (Adam Buchsbaum) (03/13/85)
Storing messages by message id in /usr/spool/news will fail horribly, because there is just too much news around. The directory would get huge, and simple directory searches will take forever.
guy@rlgvax.UUCP (Guy Harris) (03/14/85)
> Storing messages by message id in /usr/spool/news will fail > horribly, because there is just too much news around. The > directory would get huge, and simple directory searches > will take forever. A commonly used technique for keeping directory sizes down is to use several subdirectories; some function of the message ID would select the directory. This can be extended to an N-level tree if you have N independent hashing functions. Guy Harris {seismo,ihnp4,allegra}!rlgvax!guy -- Guy Harris {seismo,ihnp4,allegra}!rlgvax!guy