Package: netbase Version: 4.46 Severity: normal Hi,
For somewhat unrelated reasons I went looking today to see if there was actually a formal definition for what constitutes a valid service name ... and it turns out, that as of last month, there actually is one - that has now been published as the BCP track RFC 6335. Details of the allowable form are here: http://tools.ietf.org/html/rfc6335#section-5.1 The proposed mechanism for undoing the mess of leaving this undefined for so long are here: http://tools.ietf.org/html/rfc6335#section-10.1 It looks like the services file we are distributing should not be a major pain to bring into line with this. We have a few entries with underscores and there is: clc-build-daemon 8990/tcp # Common lisp build daemon Which fails the length constraint -- and according to IANA, that port is actually allocated to http-wmap, which hopefully makes resolving it a relatively trivial matter too :) Another issue that pops out at me here is that section 10. specifically notes that IANA consider the correct service name for port 80 to be "http" and that we shouldn't use "www" as we presently do ... so there may be other services where we aren't aligned with IANA that probably should be now too. It's not entirely clear to me what else we are going to break by changing this, but now that it's formally defined, and while we're still relatively early in the release cycle, I guess it's time to find out ... Just The Messenger -ingly Sorry, Ron -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected]
