FeedBlendr.com

I’ve (finally) moved the FeedBlendr Blog feed to being served from the wonderful Feedburner, so please update your feed URLs to point to http://feeds.feedburner.com/FeedBlendr. Thanks!

I was checking out Digg this morning and noticed this story over at Valleywag:

How To Get all the tech news you need (in 20 minutes a day)

A quick look told me there’s a better way, so here it is. The article suggests a number of news sources that you should scan on a daily basis to ensure you’re getting a good coverage of what’s going on. Checking multiple sites takes way too long, so why not blend those news sources into a single feed and then subscribe to it in your newsreader, or even view it online?

1. DailyTech: Top Sources
2. DailyTech: Commentary
3. DailyTech: Mainstream
4. DailyTech: Mainstays

The only news source I couldn’t include in there was Wired Magazine in “Mainstays”, but that’s much sexier to read in old-school hardcopy anyway, so go subscribe to it and read it on your commute or something.

These blends will all be upgraded with the new beta version of FeedBlendr soon, and then you’ll be able to access the contents of them online, on your cell-phone, wherever you want; care of some of the cool new features about to be released.

Rather than do the trendy, Web 2.0 thing and make FeedBlendr perpertually beta, I’m going to have a limited-time version up at http://beta.feedblendr.com to test out a new version of the core engine that I’ve been working on over the past few months. This is live right now, so please check it out. New features you may (or may not) notice are:

• Updated funky design (thanks again Ray Hernandez!)
• Supports podcasts/video blogs now!
• Fancy online reader for viewing the contents of your blend, including embedded audio/video
• Mobile reader – allows you to view a blend in a lightweight format specially designed for phones/PDAs etc.
• JavaScript “export” to allow you to drop the contents of a blend into your own page and style it to suit your layout using simple CSS.
• Supports custom XML-extensions (like MediaRSS, GeoRSS etc) on the specific entries sourced from feeds which contained them
• Default output format is now Atom
• Much-improved OPML/bulk feed blending operations
• Fancy AJAX feed checking that actually works
• Much better caching/crawling engine (hopefully) to improve response times and server performance
• Whiz-bang REST/OPML/XML-based API that will allow developers to build some cool stuff with Blendr as part of it (if you’re into that sort of thing)

I’d really love it if people could try out some blends on the beta version, and let me know how it goes. Specifically I want to know about any problems you have, but feel free to lavish me with praise as well 🙂 No need to tell me that the “about”, “blog”, “tips” and “contact” links don’t work either – they will when it goes live!
WARNING: Blends created on beta.feedblendr.com will NOT be transferred over to the live version once the beta phase is complete (within a few weeks), so please only use it for testing!

Having some sort of problem with refreshing the contents of feeds right now which I’m looking into.

Basically this will mean that your blends will appear to not have any new posts in them, even though the source feeds are updating.

It looks like this is actually a file-system problem being caused by there being too many files in a single directory and thus confusing the system into thinking that files are fresh (there’s over 45,000 files in cache directory…)

I’ll be fixing this today, but the change won’t be live until tonight most likely. Sorry for the delay.

UPDATE (7:00pm WST): This problem should be “fixed” now, at least for long enough to roll out the new version of FeedBlendr. The problem was indeed in the caching model, and it goes a little something like this:

MagpieRSS uses an md5 hash of the URL where it got a feed from to create a filename to store its cached copy under. That’s fine under normal circumstances, but I had over 45,000 cache files in a single directory, and PHP was choking when trying to figure out if a file was fresh or stale, and for some reason it was assuming it was fresh.

I’ve adjusted the way that Magpie caches files, so now it’s using a directory structure 5 levels deep, based on the md5 name of the cache file, which should mean that at the deepest level, there are less files per directory, thus avoiding this problem for a while.

Thanks for your patience and I hope it didn’t cause too many problems for you all out there. Let’s all look forward to the new version (coming soon, honest!), which will hopefully work around things like this!

From the suggestions of a few users, and my own desire to improve the functionality of FeedBlendr, I’m hoping to increase the number of feed formats/variants/dialects supported in the near future. At the moment, the way that FeedBlendr works means that it only supports (or rather only outputs) pretty bare-bones feeds. This means it almost always works, but it also precludes it from doing really cool things like creating belnded podcasts or vlog feeds amongst other things.

I’m going to be working from Niall Kennedy’s list of feed specs, and hope to ensure support for these formats (and possibly more if I can figure out a reliable way to open up the elements I pass through):

• RSS 0.9, 0.91, 1.0, 2.0 with enclosures
• MediaRSS
• iTunes formatted podcasting feeds
• GeoRSS

If there are any other specific formats that you’d like to see FeedBlendr support, please use the comments to let me know!

Change of plans!

Thanks to a new policy at DreamHost, there’s no immediate need to relocate, so FeedBlendr will be staying right where it is for now.

I’m hoping to get a bit of a chance to do some refactoring and what-not in the near future though, which will make blendr a little more flexible, and a little more standards compliant 🙂

Some time over the next week, I will be re-locating FeedBlendr.com to a new home at hosting.com. During that process, you may experience some problems accessing blends, but I hope to minimize those problems as much as possible.

I will post again when things look stable 🙂

With increasing popularity come a number of problems that are only to be expected. Right now, FeedBlendr is experiencing one of those problems – scalability/load issues.

After a number of warnings/notifications from DreamHost, I’ve been asked to either figure out a way to lower my processor usage immediately, or risk having my account disabled until I can do so. I’m using too much of the processor on my shared server and it’s unfair to the other users – fair enough.

Now I’m faced with a decision: do I shut FeedBlendr down (I don’t want to)? is there something I can do to lower usage (not that I’ve found yet)? can I justify upgrading my hosting (without making any money from FeedBlendr under the current model)?
At the moment I’m considering upgrading to a dedicated server from Hosting.com, which should give me the power/flexibility I need, but it’s a lot of extra management/set up etc that I’d have to deal with just to get it all happening, and life is just plain busy right now. So as a stop-gap, and either way, I’ve decided to open up for donations, allowing anyone who would like to do so to show their appreciation for the service FeedBlendr offers, by dropping a few bucks in my virtual tip-jar.

This may well affect my decision on whether or not to keep FeedBlendr live – if people aren’t even willing to drop a few dollars once-off in appreciation of the service, then I don’t know if I can continue providing it, and pay extra for the privelege of keeping FeedBlendr public and free.

It’s up to you people – so please show how you feel about FeedBlendr and donate now!

Steve Rubel has a great post over on MicroPersuasion on using FeedBlendr to track a group’s attention data. If you work with a group of RSS-ophiles (or just are one yourself), then you should check out his technique combining FeedBlendr and FeedBurner.

After a few emails from some FeedBlendr users pointed it out to me (thanks Roberta and Eric), I’ve fixed some problems on the OPML import tool on FB. You should now be able to import your blogroll from other online services such as Bloglines, OPMLManager.com and Blogrolling.com.

If there are any more problems with the service, please let me know and I’ll take care of them!