Re: linux news project impressions

[Nils Lohner <lohner@typhoon.icd.teradyne.com>]

In message <199808222353.TAA15773@belegost.mit.edu>, Roger Dingledine writes:
>In message <199808221820.OAA13864@cran.mit.edu>, corbet@eklektix.com writes:
>>It seems overly complicated and overengineered.  Do we really need NNTP
>>servers?  Do we really need XML?  If we use XML, who will put all of the
>>news items into that format?  The submitters?  If we want people to
>>actually use this service to submit news items, it has to be trivially easy
>>for them to do.  We can not expect them all to learn XML.
>
>We do not really need NNTP servers. I agree that that is an overly
>complicated solution, and I've put that part of the project on hold until
>I/we come up with a better way to handle the news transport element. As for
>XML, no, we don't *really* need it, but we need something like it. We need
>some sort of formatting language to make news portable and easy to read from
>a variety of different ways.
>
  This solution may be overly complex, but I think this system has a large 
potential for growth so this might not be as over engineered as it seems.  
Plus, NNTP is a well defined and common transfer protocol, which means that 
less effort may be required that starting something from scratch.



>
>Using a web gateway for the news means that there's a simple web form to
>fill out, with each field already listed. The web form will automatically
>convert the submission to xml in the background, and then people can convert
>that xml into whatever format of news they want to show the public, sorted
>or categorized by any of the fields.
>
  Exactly... the XML is completely invisible to the end user... it's only 
how the news is formatted once submitted, and gets decoded form there.  
Again, you might consider this overkill, but a) it does what we need, and b) 
it's been around for a while and is extensible.


>>Here is what I would propose to solve the initial problem:
>>
>>- Create a mailing list to which people can submit news items.  I would
>>  happily host such a list, with as many of you as want as moderators.  Or
>>  it could go elsewhere.  We could consider multiple lists for a
>>  first-level categorization if we want.  The list(s) would be intended for
>>  Linux news providers, but I see no reason to limit subscriptions in any
>>  way.
>>
>>- Maybe make a nice web interface somewhere for submitters who prefer that
>>  approach.  
>>
>>- Publicise the list/web site.
>
>I don't think this is any better than cola. I think it's mighty similar,
>in fact. My current plan was to
>- create a web interface for submissions, including designing what fields
>  and categories we need
>- write some basic xml->news converters. Since xml is formed like html, I
>  can use sdoc, which is a generic extensive html parser with some really
>  handy features.
>- advertise it.
>
>I'll grant you that this is more work than what you propose. It's a very
>interesting point, though, that *users* aren't going to be able to tell
>the difference if we're using xml or whatever as a backend, so it might be
>a good plan to get this set up and advertised now, and then we can fill
>in some new fields and a backend later on.
>
  Users won't be able to tell, but once it picks up and you end up patching 
whatever system you decide to use in the meantime, you'll be able to tell :) 
 And on top of that you need to do the conversion at some point anyway.  If 
you want to get it off the ground, then do it with a limited set of XML tags 
and don't implement all of the features, but have the basic skeleton in 
place when you start, otherwise you'll be doing double the work.


>>jon
>>
>>Jonathan Corbet, Eklektix, Inc.
>>corbet@eklektix.com
>