- From: Peter Koch <pk@TechFak.Uni-Bielefeld.DE>
- Date: Sat, 25 Feb 1995 16:57:01 +0100
- To: uri@bunyip.com
Reed Wade wrote: > 1. ability to specify alternate port, how important is this? There are probably no "finger" servers listening on ports other than 79, if you think of "finger" providing information about remote users. The example > finger:therm.netlib.org:451 certainly has different semantics since it acts like "send me a CRLF and I'll tell you some number and temperature unit". That is, just the kind of input and the type and format (line-oriented ASCII) of output is specified by the URL. I think, this is the right way. With ability of specifying an alternate port, WHOIS-servers could be refereced by finger://whois.whatever.org:43/Joe%20User, for example. Otherwise we had to come up with Yet Another URL Scheme for whois: . > 2. adherence to the scheme://host/ method of indicating the host > name. I can't think of a better exception to this rule than > finger--but are there any hidden costs? I would prefer finger://host[:port]/<local-part>, because I do not think there is any reason for resembling the "command line" appearance of the "finger"-command (as was argued before). Secondly, I understand RFC 1738 to provide the <user>@ part for services and/or access on behalf (resp. with identity) of <user>, as are telnet and ftp. This is not the case for "anonymous" access methods like gopher etc., and "finger" also falls into this category. The "local part" being information *about* a user is just coincidental. -Peter -- Peter Koch Internet: pk@TechFak.Uni-Bielefeld.DE Faculty of Technology Universitaet Bielefeld Tel: +49 521 106 2924 D-33501 Bielefeld / Germany Fax: +49 521 106 2962
Received on Saturday, 25 February 1995 10:57:27 UTC