Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!samsung!zaphod.mps.ohio-state.edu!lavaca.uh.edu!uhnix1!sugar!ficc!peter From: peter@ficc.ferranti.com (Peter da Silva) Newsgroups: alt.sources.d Subject: Re: wp2latex (1 of 4) Message-ID: Date: 13 Aug 90 22:16:39 GMT References: <-Z45SQ2@ggpc2.ferranti.com> <6462@helios.ee.lbl.gov> Reply-To: peter@ficc.ferranti.com (Peter da Silva) Organization: Xenix Support, FICC Lines: 28 In article <6462@helios.ee.lbl.gov> Jef Poskanzer writes: > In the referenced message, peter@ficc.ferranti.com (Peter da Silva) wrote: > }And after all that it doesn't get archived because news guys don't have the > }time to read and file every sources posting that comes in. > As I pointed out recently, you're wrong about this. The only archiving > mechanism necessary for alt.sources is compressed numbered article files > with an index of subjects. Well, yes. And a flat file system with numbered files is quite adequate for any other purpose as well, if you keep a good list of files up to date in a conventional numbered file. In fact, why use files? Just store data in numbered blocks and keep an indexed list of block ranges, like Forth does. Exxagerated, yes. But named files are a lot more convenient. > And indeed, the only operating FTPable archive > of alt.sources uses precisely this mechanism. And the BBS that I run that offers this service uses named files, because it's more convenient that all archive sites have the same name for a given file. When there's more than one FTPable archive of alt.sources how will you refer to a given article? Hey, check out 4278 on well.sf.ca.us! What's that on cs.vu.nl? Um, I think it's 23Jul90.1475@ai.mit.edu at cs.vu.nl. What do they call their index? We do keep an index file with arbitrary file names, but it's a real pain for our users to download and scan. Not everyone runs at T1 speeds. -- NAME: Peter da Silva. WOLF: `-_-' PSTN: +1 713 274 5180. 'U` DNS: peter@ferranti.com BITNET: Oavzgsz@AcadiaU.CA