how to stuff HTML into RSS??

[lkrubner]

Me and some friends are working on some PHP based templates for web pages. We've templates that look like this (simplified):

<html>
<head>
<title>
The green and blue design for carpentry companies
</title>
</head>
<body>
<?php showMainContent; ?>
<div style="width:200px; float:right">
<?php showLinkArea(3); ?>
</div>
</body>
</html>

I'd like to publish all the templates in our database in an RSS feed so it will be easier to import them on other sites. Does it screw things up if I stuff HTML into the DESCRIPTION tag on an RSS .91 feed?

 

[Andy Dingley]

Does it screw things
up if I stuff HTML into the DESCRIPTION tag on an RSS .91 feed?

It's not what you stuff, it's how you stuff it.

You should encode HTML, so that

<description><p>Some <b>HTML</b> in RSS</description>

becomes this

<description>&lt;p&gt;Some &lt;b&gt;HTML&lt;/b&gt; in
RSS&lt;/p&gt;</description>

Watch out as well for & (becomes &amp and for &eacute; etc. (turn them into the equivalent numeric entity)

I'd also suggest that you make your HTML fragments into well-formed, balanced XHTML fragments before you embed them (lower case element names, close open elements). Although this isn't required, it can make life easier with XML toolsets.

This stuff isn't hard to do, but it's very poorly documented. There are many RSS versions, and few of them describe it fully. This is a useful read http://diveintomark.org/archives/2004/02/04/incompatible-rss

I'd also avoid the obsolete RSS 0.91 in favour of RSS 1.0 (far better), or you might prefer the more popular RSS 2.0

 

[Joris Gillis]

You should encode HTML, so that

<description><p>Some <b>HTML</b> in RSS</description>

becomes this

<description>&lt;p&gt;Some &lt;b&gt;HTML&lt;/b&gt; in
RSS&lt;/p&gt;</description>

Hi,

I don't know anything about RSS, but wouldn'it be easier and more logical to insert the XHTML as elements using namespaces? And if that wouldn't be possible yet, shouldn't it become possible?

regards,

 

[Andy Dingley]

I don't know anything about RSS,

I suggest you read the Dive Into Mark article. It explains some of the background to this and is a good explanation. http://diveintomark.org/archives/2004/02/04/incompatible-rss

RSS has suffered because of too many standards, and especially because these standards have generally been poorly specified. In particular there is no clear guidance on how to embed HTML content within an RSS item.

A problem with RSS, and all such protocols that try to become an open publication medium, is that many creators will make content and many consumers will try to read it. Where the spec isn't exhaustive on how it _must_ be done, then a situation soon develops of de facto behaviour for how it _is_ done. Readers become dependent on this, and you diverge from it at your peril.

but wouldn'it be easier and more logical to insert the XHTML as elements using namespaces?

That's an attractive option. However it's not a viable one.
There are several reasons:

Namespacing relies on using XHTML, and you may wish to include HTML _as_HTML_ not XHTML. Some consumers may be confused if they receive XHTML

Namespacing relies on including a balanced fragment (i.e. one that can be well-formed as as XML fragment). This wasn't a requirement on the original RSS/HTML enclosure, so this is hard to re-impose in some cases (<a name="..." > is one of the more awkward cases to deal with).

RSS is not an XML protocol. Successive versions of badly-written specs have clouded this. There are all sorts of references of "ASCII" when it should really be CDATA. It's commonplace to include HTML entities, even when these aren't valid outside the HTML DTD. Reliable parsing of RSS from external sources is a mess, and it often relies on knife-and-fork parsing with non-XML tools. It's not reliable to assume good support for standard XML features if you're working with external feeds, even though you "should" be able to do this.

And if that wouldn't be possible yet, shouldn't it become possible?

RSS is old. It's post-XML, but pre-XHTML and (arguably) pre-namespacing. So even if a namespaced approach became widespread, consumers should (strongly) keep supporting the old way if they still want to accept content supplied that way.

I use namespaced content for internal RSS feeds within my projects, where I always use RSS 1.0. For external work though, I encode plain HTML. I use balanced fragments, so I close elements like <p>...</p>, but I don't use the <br /> form for <br>

 

[Joris Gillis]

I suggest you read the Dive Into Mark article. It explains some of the background to this and is a good explanation. http://diveintomark.org/archives/2004/02/04/incompatible-rss

RSS has suffered because of too many standards, and especially because these standards have generally been poorly specified. In particular there is no clear guidance on how to embed HTML content within an RSS item.

A problem with RSS, and all such protocols that try to become an open publication medium, is that many creators will make content and many consumers will try to read it. Where the spec isn't exhaustive on how it _must_ be done, then a situation soon develops of de facto behaviour for how it _is_ done. Readers become dependent on this, and you diverge from it at your peril.

That's an attractive option. However it's not a viable one.
There are several reasons:

Namespacing relies on using XHTML, and you may wish to include HTML _as_HTML_ not XHTML. Some consumers may be confused if they receive XHTML

Namespacing relies on including a balanced fragment (i.e. one that can be well-formed as as XML fragment). This wasn't a requirement on the original RSS/HTML enclosure, so this is hard to re-impose in some cases (<a name="..." > is one of the more awkward cases to deal with).

RSS is not an XML protocol. Successive versions of badly-written specs have clouded this. There are all sorts of references of "ASCII" when it should really be CDATA. It's commonplace to include HTML entities, even when these aren't valid outside the HTML DTD. Reliable parsing of RSS from external sources is a mess, and it often relies on knife-and-fork parsing with non-XML tools. It's not reliable to assume good support for standard XML features if you're working with external feeds, even though you "should" be able to do this.

RSS is old. It's post-XML, but pre-XHTML and (arguably) pre-namespacing. So even if a namespaced approach became widespread, consumers should (strongly) keep supporting the old way if they still want to accept content supplied that way.

I use namespaced content for internal RSS feeds within my projects, where I always use RSS 1.0. For external work though, I encode plain HTML. I use balanced fragments, so I close elements like <p>...</p>, but I don't use the <br /> form for <br>

Now that what I call a valuable reply
Thank you very much.

 

[Peter Flynn]

Me and some friends are working on some PHP based templates for web pages. We've templates that look like this (simplified):

<html>
<head>
<title>
The green and blue design for carpentry companies
</title>
</head>
<body>
<?php showMainContent; ?>
<div style="width:200px; float:right">
<?php showLinkArea(3); ?>
</div>
</body>
</html>

I'd like to publish all the templates in our database in an RSS feed so it will be easier to import them on other sites. Does it screw things up if I stuff HTML into the DESCRIPTION tag on an RSS .91 feed?

Yes. Implementations of RSS readers are almost all hopelessly broken and non-conformant, and the RSS "spec" -- such as it is -- has been so kicked about and bastardised as to be virtually worthless except as a carrier format like HTML. There were plans to make a newer, better version, but like HTML it has now become so fossilised that it's not worth changing.

///Peter

 

[lkrubner]

Thank you for your in-depth reply. I've already read Mark's article and one thing I got from it was that it didn't matter much which version of RSS you used, they were all broken.

For now I'm in the lucky position of being the consumer of my own output. We have some HTML templates we'd like publish, but we are publishing them for people who have our software, so we control the source and the point of consumption. I'd love to eventualy use a richer RSS but I'm short on time this month and so I'd like to reuse what PHP code we already have written and tested. The code we have puts out valid RSS .91.

To publish an HTML template in the description tag of RSS, should I just wrap it in a CDATA tag? Or escape it as someone ablove remarked.