Path: utzoo!utgpu!news-server.csri.toronto.edu!rpi!think.com!spool.mu.edu!uunet!mcsun!ukc!slxsys!ibmpcug!mantis!mathew From: mathew@mantis.co.uk (CNEWS MUST DIE!) Newsgroups: news.software.b Subject: Re: CNEWS MUST DIE! Message-ID: Date: 21 May 91 13:33:37 GMT Article-I.D.: mantis.q7Ta325w164w References: Organization: Mantis Consultants, Cambridge. UK. Lines: 44 brendan@cs.widener.edu (Brendan Kehoe) writes: > mathew@mantis.co.uk wrote: > >If a system falls over because of errors in its input, then that system > >should be fixed. > > Right ... but CNews doesn't have that problem. It does *exactly* what > the code intends it to do. You can throw anything at it, and it will > react accordingly, and not "blow up". I know that. However, one of the justifications given for C News's antisocial behaviour is that there are other programs which "blow up" on bad input. I am saying that those other programs should be fixed; C News should not be made to try and eliminate bad headers in order to nanny the faulty programs. > But if a good number of those articles get dropped, the sysadmin at > the receiving site will notice and (hopefully) tell the other site of > the problem. ^^^^^^^^^ "Hopefully" isn't good enough. Nobody told me for weeks, if not months. > > The sorts of errors C News drops articles for are neither > >uncorrectable, nor are they errors which cause the article to lack essential > >information. > > They are correctable only when the system can attempt to make an > 'educated' guess -- but what if it's wrong? So what if a Date: header is wrong by up to a few days? > To use that argument is to say that sending mail to foo!bar.com is > LEGAL under RFC822 because sendmail should be "intelligent" enough to > replace the ! with a @, since the local site's not connected to a > system called foo, and therefore it's a typo. Right? Wrong. In that case, I would say that sendmail should tell the user about his mistake. It does, does it not? Whereas C News doesn't. Which is wrong. mathew