Netscape Says No RSS 0.91 For You

beat.bolli asks: "As of today, Netscape has 'updated' its my.netscape.com personalized portal to version 2. It seems that they decided to drop all external RDF channels. What gives?" Well, Will Sargent writes: "Netscape removed the RSS 0.91 DTD from their website. This means that all RSS feeds which depend on the RSS 0.91 (many, MANY news sites) cannot be used with a validating parser. Rael Dornfest has more details."

Re:Alternatives? (aka SSP)

well, this may be a bit premature, but a while back, a friend and i started building our own little uber-portal, which was in concept similar to my.netscape.com, but a little bit better in implementation. (or at least we think so)

it's up and running at www.fyuze.com [fyuze.com] and for the most part, it works. the documentation is still a little confusing, and there are a few bugs, but hey, it's a work in progress.

it supports RDF, HTML and in the future XML & XSLT for digestion to other platforms (ie WAP). it's essentially free, and built on open source technology.

right now, the content selection is rather sparse, so if you run a site, add your channel/feed to the system.

at any rate, give it try. it may be a good place for all those left homeless by my.netscape.com.

(disclaimer: there is mention of charging for content distribution on the site. this is something we are still working on for high-load super-customizeable feeds. all the basic services are free to both distributor and consumer)

Attn: Submoronic Moderators

The problem isn't because there's some sudden inability to find a copy of the DTD, the problem is that there are a lot of pointers pointing to a specific place on Netscape's site. And now those pointers are broken.

I waited to see how long it would take you idiots to moderate the parent up to 5. At least one person had the sense to moderate it as "Overrated." But, even though the current score is 4, I'm ending the contest now since there've been 5 "Informative" moderations to the post.

It's things like this which causes the outside world to have the impression that the only knowledge that Slashdot users have is the knowledge of how to bitch about things that they know nothing about. And, well, I guess it's pretty hard to disagree with them.

So, thanks, you imbeciles. I'll be posting this anonymously, because I understand that you're not just clueless, you're childishly vindictive.

Conversion Tool from DTD to XML Schema

here [w3.org]

Re:RSS 0.92 is compatible with RSS 0.91

Pot - kettle - black.

RSS 0.91 was completely incompatible with 0.9; the authors of 0.91 drank Dave's Kool-aid, and made it non-RDF. 1.0 *is* backwards-compatible with 0.9, just not the 0.9x orphans.

Opera Left in the Cold

I don't expect Netscape to help out their competitors much, but I can't even get into my.netscape.com with Opera 5.0. I find it funny that even IE is allowed in, but not Opera. This is the kind of callous, inconsiderate web design approach I despise. Only allowing the main browsers, and blocking everything else, no matter how capable it might be, or what the user's choice. It's HTML for god's sake, even Lynx could view it in some form.

C'mon Netscape, kick IE out too. You've already blocked one high-quality, fast performance browser. Why not block another?

Forking RSS

I hate to say it, but this is YA example of why forking over petty differences can be A Bad Thing. If one stubborn contingent hadn't steadfastly clung onto the deprecated RSS 0.91 spec instead of moving to RSS 1.0 [purl.org] (which returns to RSS's more dynamic roots), said contingent wouldn't be locked out because *a single document was removed from a web site*. Yeah, Netscape did the wrong thing. But the proponents of the outdated and outmoded spec should have seen this coming a mile away.

No, No, No. Someone misunderstood the concept here

There is nothing really wrong with Netscape removing the DTD from the "URL" which is declared in a RSS-Document.

By Specification this URI does not need to contain a valid URL, its just a Namespace to have a unique unifier for this specific DTD.

For example take a look at the URL supplied with XSL(T)-Stylesheets: http://www.w3.org/1999/XSL/Transform. This won't give you the DTD for XSL(T)-Documents. All you get is a webpage which simply states 'This is the XSLT namespace'. And even W3C put this webpage there only after having to much people complain that this URL was in the first place 404 (not existant).

You even do not need to use a HTTP-URI for the namespace. You are perfectly legal to use a (unique!) email address or something else.

So no validating parser should do anything with this Namespace URI. At least not try to obtain the DTD from there (imagine a validating parser sending you an email asking for the DTD...). If the DTD is needed, it has to be supplied and associated by other means. Not by the URI.

So there is nothing really wrong with netscape to remove the DTD. There are copies of it floating arround the net, and they can be used for validating parsers. And for simply _using_ RSS you won't even need the DTD. Validation costs time and in most real, life application you want to turn it off anyway (if implemented at all).

Alternatives?

The My Netscape portal was great for having very customized content, i.e., anyone could create a channel and publish it. The top news stories for a lot of the sites I visit (slashdot, freshmeat, mozilla, mozillazine, and a few others) were on my front page.

Now that the custom channels are gone, what alternatives do we have for this portal?

Re:No, XML and parsers are broken as designed

DTD's are nothing like namespaces, and your conclusion is incorrect.

There is a PUBLIC identifier allowed in the XML DOCTYPE which is purely symbolic, but that's not what we're talking about here.

Yes, you misunderstood.

We're not talking about namespaces here. We're talking about DTDs, particularly the SYSTEM identifier, which is meant to be fetchable (it even says so in the XML spec).

RSS 0.91 isn't even namespace aware.

Plus, you say "for simply _using_ RSS you won't even need the DTD". That's only *mostly* true, unfortunately. The RSS 0.91 DTD contains all the HTML entity references that are commonly used in RSS 0.91 files, and an XML parser will choke without these being defined somewhere (and that somewhere happens to be the DTD) (well actually a non validating parser can just ignore the entities (providing they are syntactically correct), but then you'd end up with strange looking content).

Re:Yes, you misunderstood.

Once again: To use a validating parser you need the DTD, of course. But the W3C makes no provision where to get this DTD! The Declaration inside a RDF-File for example is containing a xmlns-Attribute. That is XML-_N_ame_s_pace. Nothing more. See for example the declaration in http://slashdot.org/slashdot.rdf:

OK:

$ g-request http://slashdot.org/slashdot.rdf
<?xml version="1.0" encoding="ISO-8859-1"?><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax -ns#"
xmlns="http://my.netscape.com/rdf/simple/0.9/">
...
Sorry, no cigar. This does not contain a DOCTYPE declaration. Try an RSS 0.91 file, e.g. from xmlhack.com:

$ g-request http://xmlhack.com/rsscat.php
<?xml version="1.0" encoding="ISO-8859-1"?>
<!DOCTYPE rss PUBLIC "-//Netscape Communications//DTD RSS 0.91//EN"
"http://my.netscape.com/publish/formats/rss-0.91.d td">
<rss version="0.91">
...

You can not run a validating parser solely on this. Because the URIs given above are in fact (what you appearently do not want to understand) only Namespace, and not a reference to the DTD.

No, you cannot run a validating parser solely on the first one because it uses XML namespaces! Try the RSS 0.91 file format. That's what we're talking about here.

There is no defined way to automatically determine where to get the DTD. You need to get it yourself and supply it to your validating Parser.

I, and the XML spec, beg to differ. I quote:

[Definition: The SystemLiteral is called the entity's system identifier. It is a URI reference (as defined in [IETF RFC 2396], updated by [IETF RFC 2732]), meant to be dereferenced to obtain input for the XML processor to construct the entity's replacement text.]

How do you dereference HTTP URI's? Well shock horror you use the HTTP protocol! ;-)

I've given the XSL(T)-Example. The XSL(T)-Declaration contains a URI as well.

No, it does NOT. There is no Declaration, only a start element. The reason being that XSLT uses XML namespaces, and namespaces aren't compatible with DTDs (well, they *mostly* aren't compatible - but I'll leave it to a book like XML in a Nutshell to explain the finer details).

But the DTD is not there, only a information by the W3C, that this is the XSL(T)-Namespace. After all you may imagine that the W3C knows what it does.

Correct, the W3C does know what it does. The document pointed at by the namespace (which as you correctly, but confusingly point out, doesn't actually need to exist) at the W3C site is in RDDL. Go to XML.com to find out what RDDL is. (actually it might not yet be in RDDL, because the W3C is only slowly moving over to RDDL. But the Schema namespace *does* point to a RDDL document, and this is a sign of things to come).

A valid XSL(T)-Document does not contain any reference about where to get the DTD.

That's because XSLT can't use DTDs, because it uses XML namespaces.

This was specifically intended so, as the Locations of DTD can change, as the example here shows us.

Actually the reason namespaces chose not to have anything at the URI is because they didn't have a namespace compatible schema system in place at the time, and so nobody could agree what to put there. Now the people on the XML-Dev mailing list have agreed on putting RDDL documents there.

So you are still wrong. Having the DTD disappeared from Netscapes site DOES NOT make problems to validating parsers by default. Just obtain the DTD elsewhere and feed it into your parser.

That's possible if you implement a custom URI resolver, but barely anybody does that because it's not the right way to go about fixing the problem (which is to use a catalog system based on the PUBLIC identifier instead).

But this IS NOT A PROBLEM BY CONCEPT. The W3C recognized that it can be virtually impossible to maintain a ever lasting repository for a DTD, and therefore does not urge you to make any links to your DTD. The URI in the Declaration is only Namespace, and you don't even need to put a Webpage there. The W3C itself had not webpages at there URIs at the beginning. Only after having too much people who do not understand (like you) complaing the added this short text. You can use mailto:bla@fasel or gopher:.. or telnet: or anyother unique URI you might imagine. Just make it unique for a given DTD.

You're talking about namespaces again, not DTD's. Please go and buy a good book on XML (such as XML in a Nutshell, which I was one of two tech editors for) that will help clear up your confusion.

Re:Opera Left in the Cold

Frankly, I can't imagine why everyone doesn't just say "fuck Netscape," and start using a *good* browser.

Every since their debacle with having valid CSS cause hard crashes to the browser, I, for one, refuse to code work-arounds for its HTML and CSS deficencies.

Choose Better: Choose Opera.


--

Web vs Paper publishing

This highlights one problem from which Web publishing suffers compared to (tradional) Paper publishing. If a reference is made to a Web document (via URI) then if the publisher removes or changes the document the reference becomes invalid, or refers to the "wrong" thing. When the reference is made to a paper document, the book may go out of print (which in the case of journals is almost immediate), but this does not prevent you refering to the copy you already own or visiting the library to consult it.

Re:For those who don't know: What is RSS?

Dave has also put the DTD back up on one of Userland's site, available at:

http://www.scripting.com/dtd/rss-0_91.dtd [scripting.com]
---

Re:No, XML and parsers are broken as designed

Depends how much you want to code. Certainly the java parsers I've used (xerces etc) let you do entity resolution yourself and so look for the dtd anywhere you want to. In fact I've actually done this.

Re:Corporate justification of mistrust..

It's my understanding that there are two camps here:

1. Those who believe that the DTD URI is a real live URL to be fetched to contain the DTD
2. Those who believe that the DTD URI is merely a URI, to distinguish that DTD from all other possible DTDs in the world.

All AOL-Netscape did was mess with the heads of the first camp. Those who were in the second camp are still chugging away fine.

It's also my understanding that those who developed XML are firmly in the second camp. Yes, a URI-to-URL mapping mechanism needs to be developed for DTD URIs. But just because that's not in place, let's not rank on Netscape for their choice to make the one-to-one trivial mapping no longer valid for RSS 0.91.

Re:For those who don't know: What is RSS?

bzzzt! wrong -

RSS stands for RDF Site Summary, not Rich Site Summary. And RDF, as we all should know stands for Resource Description Framework.

So, RSS really expands to Resource Description Framework Site Summary.

No wonder PCMCIA.

Re:It's all about the user agent

<script>
var ua=navigator.userAgent;
//redirect badbrowsers
if (ua.indexOf('MSIE 4')!=-1||ua.indexOf('MSIE 5')!=-1) redirectPage=false;
else if (navigator.appName=="Netscape"&&parseInt(navigator .appVersion)>=4)
redirectPage=false;
//redirect all other clients not satisfied above
else redirectPage=true;
if (redirectPage) location.href="/shared/badbrowser.psp";
</script>

That's what's doing the browser recognition (and "bad browser" redirection)
--
Trelane -------------------------------------------------- ----------

The DTD is available elsewhere

From Dave's Scripting News [scripting.com] on Friday, 27 Apr 01 [userland.com]:

From the If-It-Weren't-So-Sad-It-Would-Be-Funny Department, yesterday when Netscape (apparently) deprecated [dictionary.com] RSS and broke all the links to their RSS stuff, they also broke people whose XML parsers require a DTD. The old URL [netscape.com] for the RSS 0.91 DTD is totally 404 not found. John Munsch has a report [yahoo.com] from the field. I put a copy [scripting.com] of the DTD into a folder here on scripting.com, and it will stay there, Murphy-willing, for perpetuity.

You can find his copy of the DTD here [scripting.com].

J.J.

Re:It's all about the user agent

I tried to visit the page in Konqueror and got an 'Invalid Browser Configuration.' (http://my.netscape.com/shared/badbrowser.psp).

These notices are showing up at various places around the web. I guess they think writing things using w3c standards is too difficult in comparison to writing them for two browsers (IE 4/5.x and Netscape 4.x).

No, XML and parsers are broken as designed

XML's flaw was using a URI as a primary key to an document (or object) type, ignoring the cost of maintaining a stable URI (especially in these days of WIPO domain name arbitration and companies that rename themselves to escape the past) and ignoring the established wisdom of never exposing a primary key to the public. The whole DTD mechanism was poorly thought-out as well; why would you ever want to carry a DTD around with a document unless you foresaw a need to validate the structure of a document by agents that don't have it? This greatly reduces the utility of validation to near uselessness.

Validating XML parsers, on the other hand, screwed the pooch by not providing for documents to be validated against any but the specified DTD. At the very least, new validating parsers ought to maintain an internal table of aliases, so that you can validate documents of type http://www.netscape.com/~nportman/grits.dtd against file:/home/marxmarv/project/dtds/hotgrits.dtd without changing the XML DTD specification or the input at all, and would also enable DTD's to be specified by, say, MIME Content-Type.

None of this fixes the copyright problem, but it does keep entire applications from going away because of some sysadmin cock-up. (As an aside, at least .NET will show the sheeple how often network blackouts happen...)

-jhp

Mozilla & My Netscape?

The new My Netscape looks great in IE, but it's bad wrong with Mozilla. What gives?

Re:RSS 0.92 is compatible with RSS 0.91

That's one thing that the RSS 1.0 fascists won't ever tell you. RSS 1.0 is NOT compatible with RSS 0.92. It's a totally different syntax.

Of course, they'll whinge and moan and say "upgrade to our new fancy spec" that only about eight geeks in the world can comprehend. When the truth is that 0.91 was a standard that EVERYONE could understand, including Frontpage-using HTML novices - EXACTLY the kind of people that the uber-geek elite want to stamp out of existence.

Sorry, geeks, it's a big wide web out there, and everyone's invited. Stop trying to bury web standards under layers of incomprehensibility.

In the meantime, as Alien54 suggests, RSS 0.91 providers can switch their Doctype to 0.92. The DTD's on Userland, which at least is run by someone with a clue.

Then everything should start working smoothly again, without having to learn an entirely new specification.

Re:Use WDDX for syndication.

Have you seen the license [openwddx.org]? This is just another way of going around the same block again. The lesson is, don't trust or support any commercial entity with Web standards, whether it be AOL, Allaire, Macromedia or any other. Trust only genuinely open community projects and genuinely open not-for-profits.

Konqi, too (was Re:Opera Left in the Cold)

Interestingly, it redirects Konqueror 2.1 [konqueror.org] to 'badbrowser.psp' [mynetscape.com]; but if you configure Konqueror to pretend to be Mozilla/4.0 (compatible; MSIE 5.5; Windows NT 4.0) it renders just fine. I'm seeing this more and more frequently with commercial sites - if they don't recognise the browser, they don't make any attempt to render a page for it. I think this is professional incompetence, frankly.

Walled Gardens

See people, AOL is the real enemy here. Within their walled garden they want you to only see AOL/Time Warner properties - "screw the rest of 'em" is now job one, more than ever. "Synergy" is one thing, this is a whole other ball o' wax.
--
OliverWillis.Com [oliverwillis.com]

Corporate justification of mistrust..

The last sentecne says it all:

RSS partisanship aside, this episode strikes yet another blow against the use of centralized (specifically copyright) DTDs in an increasingly distributed computing environment.

Publicly used DTDs need to be somewhere where the public can count on them long term.

Or else we need a DTD caching mechanism with an inifinite TTL - and this *still* doesn't address the copyright issue.

--

Actually, that was 3 days ago...

It's bugged the hell out of me. In the meantime, there's still my.userland.com, xmltree, and weblogs.com to name a few sites that still make wide use of the RSS 0.91

RSS 0.92 and onward

If you're looking for the already implemented RSS 0.92 [userland.com] look here There's also a reference to RSS 0.93 on which development started on April 01,2001.

It's all about the user agent

That is very true. It looks fine in Netscape 4.76 and MSIE 5, but it's all one wide column in in Mozilla. Using Opera [opera.com], I get varying results, depending on what "Identify As..." setting I use. Opera has the nice feature that I can set the "User Agent" string it sends to the server to be Opera, Mozilla 5.0 (which is what Mozilla / Netscape 6 sends), Mozilla 4.76 (aka Netscape 4.76), Mozilla 3.0, and MSIE 5.0.

When set to "Opera" or "Mozilla 3.0", the My Netscape page doesn't load at all, displaying an "invalid browser" page, saying you must use NS 4.x or IE 5.x. (Apparently they think NS 6 sucks too). Setting the User Agent to "Mozilla 5.0" or "Mozilla 4.76" (strange, since the real NS 4.76 works) result in the same one-columned-stretched layout as Mozilla. However, when Opera is emulating MSIE 5.0, it works perfectly!

So apparently, it's not that the page couldn't display in Mozilla, it's that My Netscape intentionally screws up the page for Mozilla!!

Re:For those who don't know: What is RSS?

Rich Site Summary (RSS 0.91) is lightweight syndication format for distributing news headlines on the web. It is a format that originated at Netscape for syndication of content through Netscape Netcenter. The format was also influenced by the the Channel Definition Format, which Microsoft defined and saw its window of opportunity open and close with "push" technologies. Userland.com built and hosted the first RSS aggregator outside of Netscape. Dave Winer of Userland formalized the Netscape specification as RSS 0.91.

Re:Hmmm...

This really sucks. How about an open source version of my.netscape.com... like a my.mozilla.org...

You seem to miss the point. A brief explenation of the problem:

The article is about RSS files, which are files used mostly by news sites and blogs. They contain a summary of the most recent topics on a site. RSS files were used by my.netscape.com to monitor multiple sites on your MyNetscape page, also they are used in Evolution and the Nautilus has the ability in CVS (under a "News" sidebar IIRC).

The problem here is that the RSS format was written in XML and used a DTD (document type definition) that was stored on the Netscape servers. Whenever *someone* *somewhere* tries to parse a RSS file the Netscape server is queried for the file and the RSS file is validated against it. So now that Netscape removed the file people don't get to see the RSS summary but get an error instead.

What could be done is putting a copy of the file on an alternate location and changing all RSS files to match the new URI... well, this could be done if it weren't for the fact that Netscape copyrighted the RSS DTD... the only sollution left is to change to the updated RSS format which doesn't depend on Netscape.

Re:No, XML and parsers are broken as designed

If DTD's are anything like namespaces (and why shouldn't they be?), the URI is entirely symbolic -- you shouldn't expect to find anything at the location specified, at least according to my copy of XML in a Nutshell.

--

Use WDDX for syndication.

Check out www.openwddx.org [openwddx.org] and move your data over to an open source alternative backed by the web giants, Macromedia/Allaire. Do the web a favor and don't support AOL.

RSS 0.92 is compatible with RSS 0.91

If you're looking for the already implemented RSS 0.92 look here [http://backend.userland.com/rss092 [userland.com]] There's also a reference to RSS 0.93 on which development started on April 01,2001.

This is merely Vital information.

As seen on the site:

How 0.92 relates to 0.91
RSS 0.92 is upward-compatible with RSS 0.91.
Every new feature of 0.92 is optional, meaning that a 0.91 file is also a valid 0.92 file.

Now if Netscrape would only document this better and let the rest of the world what is going on.

Check out the Vinny the Vampire [eplugz.com] comic strip

Re:Is there a simple solution?

Here [kde.org], by the way, is an example where some silly issues got in the way of a well-meaning project, objections were raised in a courteous way and the situation was resolved to everyone's satisfaction. Fortunately it was all cleared up before the story made it to Slashdot [slashdot.org].

Unsettling MOTD at my ISP.

Is there a simple solution?

I know it's customary here to fly into a frenzy over any perceived legal obstacle, without making any attempt to contact the people involved to see if there's a simple resolution.

But on the off chance that the 1000th time is going to be different, let me ask this -- has anyone considered just asking Netscape if they'll make the DTD public? The O'Reilly writer doesn't seem to have bothered.

As far as MyNetscape itself is concerned, good riddance as far as I care. They came up with the clever stratagem of blocking Mac IE users. (The notice that there are some unresolved problems with that browser has been there for over a year.) Good move, guys -- piss away all of your market share and then block IE users from the one thing you have left. I switched to Yahoo, with no regrets.

Unsettling MOTD at my ISP.

My my

When introduced in 1999, the "My" concept itself wasn't anything earth-shattering

What about those of us who use Perl [perl.com]?

Is Sidebar Better?

I think RDF was a good setup for a while, but Mozilla's sidebar stuff seems better off. You can use good ol HTML, make the list look better than a Unnumbered List and even add a little color. DOne right, sidebar channels work very well and could easily be included where RDF files were before. The only problem is a lack of standards so YMMV depending on teh site whose content highlights you are trying to get.

I mean an RDF file was nothing more than a <UL> list that used a bizarre, tedious format. Yes, it WAS standardized so a valid RDF channel would generally be usable. But what if it's not? Its the source sites lose - you'll just remove it.

--

My.solution

Being a type-A personality who charges off and gets to work without thinking things over, I immediately went and squatted on "dtdcache.org", intending to set up a donation-supported central repository for DTDs. This would solve problems caused by changing web structures and provide a permanent location for static XML specifications.

Of course, given an additional 30 seconds of mature reflection, I realized that XML.org [xml.org] would be the best host of such a repository, probably tied into their XML registry program [xml.org]. Ah well.

Customer Service Issue, too ...

This really is a fairly big letdown.

Another point that could be made here is over real customer service and what Netscape obviously perceives, or perhaps more importantly does not perceive, as a customer.

After all, companies are companies and, grand schemes and noble ideals aside, they're in it for the money, no question. Interestingly it seems that just in the last twelve months or so the realization of marketplace pressures, revenue streams and real product distribution, etc., have become as important as they are in any industry. It's valuable to see how the companies that are surviving in the "e-" sector, which would include Netscape even as a progenitor of that sector, are adapting. In this case the question is, "How do we identify with our customer?"

In some sectors this is not taken for granted. The automobile industry spends, quite literally, millions on trying to identify the target for an automobile down to the clothes those people buy, where the buy them and how they like to wear them, along with a myriad of other details. Sometimes, however, (to make a leap) it seems as though software companies that cater to developers occasionally just assume that since they've got developers on staff then, well, they intuitively understand their customers by a proxy of self-identity -- not always, perhaps, but seemingly more often than, say, oh lawnmower manufacturers.

It's an unfortunate truth, however, that it's easiest to hold contempt for your own kind. Someone can go through their whole day being nice to strangers, only to reserve frustration and bottled up whatever for the lucky family suffering at home with them. The same, it seems, held true for Netscape.

Marketing shmarketing. Customer shmustomer. RSS? They don't need their stinkin' RSS! Pull it! Screw 'em! Or maybe it was just a more gradual ah-who-even-cares decline. Who knows. Maybe they have some grand plan in store, something that will make up for what in any other industry would really be perceived as a Microsoft-scaled blunder. But it sure doesn't seem like anyone over there is listening to the customer. You know, those guys that are supposed to be, in some way, paying the salaries over there?

Of course things like this happen in other companies and to other customer bases, usually right before that company truly begins to flirt with really large and impressive failures. And judging from Netscape's track record of listening -- how many years went by waiting for Netscape 6 while Microsoft slowly crept upon the unsuspecting town and preyed on the needs of the poor townfolk? -- this is just another step in the wrong direction.

Interesting move, Netscape. People hate Microsoft with a passion. But people are just starting to look at you with, well, *yawn* ...