Sunday, October 30, 2005

Validating Link Rolls

For some time I've been using the del.icio.us link roll script to display links from del.cio.us, and a variety of RSS to javascript services to render arbitrary RSS feeds in my own web pages.

The time has come, however, to stop relying on 3rd party scripts, and to run my own service. A quick trawl of the web turned up several PHP scripts, but I think the one I'll opt for is Feed2JS.

However, as with most of these services, the script is lacking in what I regard as a sensible extension - link checking. After all, why display a link if it's down?

So after another quick trawl, for link checkers this time, I think I'm going to mod this unliked URL checker, which as it stands looks through a list of links extracted from a database, and writes out a list of URLs that are down. The mod will of course just search through the links provided in the link roll, and only write out those that are sound.

There's a lot more can be done to add value to RSS2JS services, I think (at the expense of server load, of course), such as taking into account permanent redirects (as suggested in this blog on link maintenace in blog posts, and which actually quotes from this more comprehensive post on the ideal blog post).

Thursday, October 27, 2005

Coping With Email: Is Hope the Only Way Out of Here?

Although I manage to get the number of messages in my primary email Mailbox down to a manageable number every so often, I find as each day goes by that the number of unread and unfiled messages tends to increase inexorably, until I somehow catch a glimpse at the number of messages (over 1500 again!) and wonder at how to get the number down without the email equivalent of my annual end-of-year "sweep whatever's on and under my desk into a binbag" trick.

At one point I read Mark Hurst's Good Experience: E-mail Management Report, Reducing E-mail Overload report, hoping it would tell me things I didn't know (and I have to admit it did prompt me to take another look at my filters), but the fixes didn't work for more than a week or two, and here I am again facing an email overflow...

So when I was pondering this on my drive home yesterday, I realised that a tool that might help me start to clear my mail box, and contribute to it's ongoing management, would be a tool that offered me hope that I was getting my mail under control.

And what is that tool? A simple display of a simple counter, or set of counts, of:

  • number of emails received today/yesterday/this week (let's call this R);
  • number of emails sent today/yesterday/this week (let's call this S);
  • number of emails deleted (D);
  • number of emails filed (F).

If in any given period (R+S) > (D+F), my Mailbox will have an increasing number of messages in it over that period. This is a Bad Thing.

If in any given period (R+S) = (D+F), my Mailbox will be in a steady state. Depending on how many messages are in the Mailbox, this may be a Good Thing or a Bad Thing.

If in any given period (R+S) < (D+F), my Mailbox will have a decreasing number of messages in it over that period. This is a Good Thing.

Colour coding the net increase or decrease of messages in my Mailbox in realtime within the mailbox interface would let me know whether there was any hope in waging a war of attrition against my email clutter, or whether it was once again on the path to overwhelming me.

I mentioned this in passing to my mailadmin, who was taken enough by the idea to consider knocking up a widget for her preferred mail client (OpenText's FirstClass). If that comes to fruition, I'll give it a whirl and let you know how I get on!

PS this was my first blog post made with Flock and it was pleasingly straightforward...apart from the fact that I chose to blog the Good Experience page - which inserted a link into the post that I couldn't easily put text in front of without it being underlined and treated as part of the link... until I right-clicked and found a Discontinue Link option! And the fact that the bullet item button didn;t seem to work properly...

Thursday, October 13, 2005

Search Rolls and Feed Annotation Streams

As well as being able to subscribe in one fell swoop to every feed listed in a given feedroll, I also like the idea of being able to call on predefined searchrolls (e.g. that aggregate a list of OpenSearch-like search engines, or correspond to a particular Rollyo search profile) as part of an aggregation profile.

Taking the idea of meta one step further, it would also be useful to syndicate the result of particular searches (i.e. searches on particular terms) over all the sites contained in a searchroll. And I wonder whether fastreams would be a way of adding the search terms to a search roll to enable such functionality?

Aggregation Profiles

With the increasing number of knowledge sources that are making themselves amenable to subscription via RSS, working out what to subscribe to to get coverage of a particular topic is likely to be increasingly problematic.

The availability of blogrolls and linkrolls are potentially exemplars of what I have been thinking of as aggregation profiles - or lists of things I can subscribe to on a particular topic.

It's easy to imagine a repository of such profiles, perhaps as an extension to Wikipedia entries, which gives users looking to subscribe to a variety of feeds in a given area (so they get an overview, or coverage, of that area) a good starting point.

One way I saw these working was to provide a user downloading an aggregation profile (or feedroll?) and offer an immediate subscription to all the feeds in that feedroll. A simple management dialogue, with the default of a ticked checkbox for every feed in the profile would allow users to customise the profile from a rich starting point.

Catching up on some unread posts today, I see the idea is in the elsewhere, with the proposal of RSS Readling Lists, which a user can subscribe to in toto, and which automatically passes through content from any, only and all the feeds contained in the reading list at any given time. (Note the element of customisaiton is not proposed in the reading list idea).

I look forward in my lazy web way to seeing the first few clients...

Wednesday, October 05, 2005

Social Developer Tools

I've been holding off blogging about Ning, even though I managed to bookmark it quite early on (which has done my del.icio.us collaborative rank no end of good!), partly because a lot of people I read have been blogging the release announcement, partly becuase I couldn't think of anything to add application, or use wise.

For those of you who havent looked yet (or had a play yet - I wonder how long before I get my developer account?;-) Ning is, in part, a tool for building social software websites. If you run your own webserver, and are happy with PHP, Perl, or Ruby on Rails, I imagine there may be an element of so what? about this (apart from the high level of respect for what the Ning folks have achieved technically.

After all, if you want a social site, you can code it from scratch, right, or at least install something you found on Sourceforge, can't you?

Well - err - no, actually... That's just not really been possibile for most people until now, unless you knew a friendly sys admin that is.

But Ning changes that - because it appears that Ning will let you clone and run existing Ning applications with a variety of social tools in the mix for free - tagging, bookmarks, photo collections, geo- tools, review sites, and so on...

So where Ning is really neat, I think, is in the way that is socialises the creation of social website construction, or at least makes it possible for mortals to create their on social websites...

Another interesting element is the meta-sociality aspect, whereby it appears that Ning sites are socially interconnected to each other via tags, etc.

I haven't yet thought through this side in much detail, but when I do it'll be in the context of elearning and distance education, and probably over on my OUseful blog.