Path: utzoo!attcan!utgpu!watmath!att!rutgers!ucsd!tut.cis.ohio-state.edu!purdue!bu-cs!xylogics!world!bzs From: bzs@world.std.com (Barry Shein) Newsgroups: news.groups Subject: Re: Give it up, folks (REALLY: THE SOLUTION) Message-ID: <1989Nov15.231248.11580@world.std.com> Date: 15 Nov 89 23:12:48 GMT References: <36339@apple.Apple.COM> <10119@stag.math.lsa.umich.edu> <36343@apple.Apple.COM> <1989Nov11.002535.21243@world.std.com> <255F14A9.16035@ateng.com> <1989Nov15.160912.8148@world.std.com> Organization: The World @ Software Tool & Die Lines: 69 Look folks, the solution to the complicated name space problem is pretty trivial. And it ain't "let's try and require 1,000 flame posts before we create a group." The current group names, like sci.aquaria (ahem) are nothing but a catalog entry, not unlike QA101.75V2 you might find on a library book (and could be better structured, but that's not what I want to say here.) All it really needs is more descriptive info. Why not something like a termcap file for groups? Two types of keywords will be mandatory and unique, the newsgroup name (sci.aquaria) and a longer, natural language oriented entry ("Aquarium Science".) To that you can add properties and other non-unique keywords so people can enter a query (fish?) and get back all the groups which match, and choose one (Matches: sci.aquaria, rec.food.cooking.sushi, choice?) It's up to the user interface whether that affects the newsrc file or not, etc etc etc. Let's call this file the newsgroupcap file and set it up like this: sci.aquaria|Aquarium Science|\ :mod:kw=fish,aquariums,gravel,guppies:\ :exp#14:dist=sci:\ :desc="A moderated group for the discussion of aquarium science":\ :mail=richard@gryphon.com: Key: mod boolean, group is moderated kw list of keywords exp expiration interval in days dist default distribution of postings desc short description of group mail mail address of moderator ...etc... Most of these entries can be standardized and distributed just like everything else. Now we can write a little set of routines to query the file just like the termcap routines. Heck, it even works with YP software, you can distribute it on your LAN trivially. User interfaces can compile it into more convenient indices (eg. by keyword.) Just like our experience with termcap (and terminfo), several dozen different little flags and variables can be added with no great confusion. It's hard to believe that this should get out of hand so let's drop that conversation. Less popular user interfaces can have their own flags added if they like since unrecognized flags are ignored anyhow by other interfaces. These may or may not be distributed as standard, that's a different issue, who cares really (given a set of "standard" flags most of you could generate the simple shell script to remove all the non-standard ones or insert new ones, really not a major concern.) Local settings (like expiration) can be managed locally (perhaps some convention is needed to mark the beginning of local flags so it's easy to recollate with updates) and files like explist can be regenerated from this master file, automatically. Same for the sys file etc. Now the user interface freaks can design to their hearts content and we can probably surivive about 100,000 groups before the system breaks down and has to be rethought out again. Done. -- -Barry Shein Software Tool & Die, Purveyors to the Trade | bzs@world.std.com 1330 Beacon St, Brookline, MA 02146, (617) 739-0202 | {xylogics,uunet}world!bzs