RE: Atom 0.2 feedback

[Ziv Caspi <zivca@xxxxxxxxxxxxxxxx>]

Tim Bray writes:
[...]

> 3. The 0.2 snapshot is actually less functional than RSS in one
> important respect.  At 'ongoing' I tend to write really long entries.
> With RSS 0.2, I can excerpt them in the RSS feed and still (via the
> horrid escaped-HTML kludge) put hyperlinks and some styling in the
> excerpt.  Since <summary> doesn't allow markup, I can't do any styling
> there, right?  So is it OK to put my teaser in <content>, which does
> allow this, or is the semantic of <content> that the content must be
> complete?

Putting a partial content and calling that <content> is the wrong thing to
do. We really should make it clear that <content> is full content, always.

The fact that your RSS 2.0 feed (not RSS 0.2! :-) has partial content
pretending it's full content is a pain for some readers. (Well, it is a pain
for me.) The reason is that the aggregator doesn't understand that it only
sees partial content, and so cannot make intelligent decisions on how to
proceed.

> At ongoing, if the <description> in the RSS is *not* the full
> text of the entry, I signal this with a trailing "...".

Relying on "..." as a visual indicator to human readers that the content is
partial can only be considered a hack. (For example, if you're an offline
reader, there's nothing more irritating than seeing those three dots on an
interesting teaser.)

What's worse, this is not even standard practice. In many feeds one (machine
or man) cannot tell by reading the content if it's partial or not.

Hey, we're in the position to fix that!

> If we are going
> to allow non-complete <content> elements should this be made explicit
> with a full=(yes|no) attribute?

I suggest we either:
1. Have a partial=(yes|no) attribute, with "yes" the default
2. Have a different element, called either <teaser> or <excerpt>, which is
similar to <content>, but indicates that more information can be retrieved
either by going through to the item's link, or (even better!) by downloading
an Atom resource from the URL indicated by <link-to-atom>, which contains an
Atom <entry> with full <content>.

> Put another way, it seems wrong that the ability to style/markup is
> coupled to whether or not you want to provide complete content.

+1

Ziv Caspi
  cell: +972-53-668-751
    web: http://radio.weblogs.com/0106548/