Re: [syndication] Re: XML-RPC and the Need to Cash In

[Morbus Iff <morbus@disobey.com>]

>For desktop apps, how about looking at the user's Web browser history
>files, to figure out whether they've visited it via some other route?
>Offhand I don't know the feasibility of doing this for common browsers,
>nor if anyone has already done this.

That would be a rather interesting idea. Lemme check on my Mac (rummages):

 IE 5.1 (mac) stores the History in bookmark format in
 ~/System Folder/Preferences/Explorer/History.html

 Netscape 4.7 (mac) uses BerkDB as well: Netscape History: Berkeley DB
 Hash file (Version 2, Big Endian, Bucket Size 4096, Bucket Shift 12,
 Directory Size 256, Segment Size 256, Segment Shift 8, Overflow Point 4,
 Last Freed 3, Max Bucket 12, High Mask 0xf, Low Mask 0x7, Fill Factor
 46, Number of Keys 120) and follows the same directory path as the
 Windows version.

 iCab I haven't yet got around to checking, but it does store a
 physical History file in the Pref's folder.

>Plausible?

It certainly sounds plausible, and would alleviate the need to have a link
tracker like Dave suggested [1]. Hey Dave, any desire to figure something
out for this?

At 3:24 PM +0000 8/26/01, Mark Paschal wrote:
>Since finding what someone's read is not only hard but not necessarily
>a good indicator of what's safe to ignore, what about not depending on
>what's been read to reduce redundancy? One way would be to present
>"similar items" (however defined) as a single group, in something like
>a "top news" section. That would make it harder to find all the

So, in essence, a locally created blogdex [2] based on currently shown
feeds? We could do something like "This [url] was referenced [x] times.
Other sites that may be of some interest would be: [other sites linked in
the same item]."

At 1:17 PM -0700 8/28/01, Mike Dierken wrote:
>I wonder if the 'visited' link style in HTML can be set to be 'hidden'.
>That >way the links in multiple categories that reference the same
>resource just go

I believe it can, but there'd have to be some logic that says "if this
item's link becomes hidden, hide the entire item itself". And I'm not sure
doing that in Javascript or any other DOM language would be a good idea.

At 1:26 AM +0200 8/29/01, selm wrote:
>Maybe new rss items can help here, like timestamps to sort by date to create
>something like a threat, multiple referres can help too, a refered link, so
>we can sort news that all link to the same story into a category..

I'm not too enthused about relying on new RSS tags to solve our problem.
Adoption of 1.0, whilst gaining, has been relatively slow (along with the
.92 and .93 versions). Even if someone added a 1.0 module that would solve
the problem, there'd be too much of a curve to get people's writer
applications to support and produce.

At 4:29 PM -0700 8/28/01, Jim Winstead wrote:
>i suspect there are corresponding dublin core elements that can be
>used with rss 1.0.

I'm no expert on DC, but from what I can tell [3], the date elements are
for the entire feed itself, not for individual items, which would be the
ideal solution.

[1] http://groups.yahoo.com/group/syndication/message/2239
[2] http://blogdex.media.mit.edu/
[3] http://webreference.com/xml/column24/3.html

-- 
Morbus Iff ( i am your scary godmother )
http://www.disobey.com/ && http://www.gamegrene.com/
please me: http://www.amazon.com/exec/obidos/wishlist/25USVJDH68554
icq: 2927491 / aim: akaMorbus / yahoo: morbus_iff / jabber.org: morbus