Last Build Date: Fri, 01 Aug 2003 09:45:06 GMTCopyright: Copyright 2003 The FuzzyStuff
Fri, 01 Aug 2003 09:44:54 GMT
I'm sorry but Wireless (Linksys specifically) on Windows can just plain bite me. Yesterday I had to work with Francois at his place so I had to reset all my wireless settings. Now this morning as I'm trying to get my email and run out the door for a plane (offline all weekend) -- nothing works. I do the restart dance. Nada. I shut down and pull the card. Nada. Zippo. Bupkus. Zero. Finally it comes up. I have to hand it to Apple -- I've never had an iBook have Wifi issues. Us Windows folks are sitting there swearing to beat the band and the Mac folk just look smug. I just wish Apple could make a decent keyboard. I'm realizing that for my next laptop, I could buy a powerbook, run Yellow Dog Linux on it and at least have a machine where ALT+TAB worked properly. And, of course, one where installing an application like QuickTime didn't require a restart. What's up with that?
Sigh. Wireless on Windows. Sucks.
Thu, 31 Jul 2003 14:15:57 GMT
Heh. If you track our Feedster Stats page at all (and yeah its slow to display and I know I need to make a cached version) then you'll see something interesting: 160,000 + feeds now. That's right. Last night we added every Live Journal user to Feedster which doubled our database size. My hat's definitely tipped to Francois whose crawler architecture happily crawled > 1,500,000 posts in 12 hours. Nice job man!
Jon Thompson nicely pointed out to me that while we might have implemented Sherlock support, we made it hard to find. Thanks man. Done! Now its linked to from the home page until it finds a permanent home.
Now for a miscellaneous selection of oddments:
Wed, 30 Jul 2003 22:26:29 GMT
Hm... We've had an increasing number of people ask us for Palm / PocketPC / Sidekick / whacky device here support in Feedster and we actually do want to address this. The real caveat here is that my current Palm is a Palm IIIx (no Internet access there), Francois doesn't even use a mobile device and since we're a teeny, tiny unfunded startup, there just isn't a hardware budget. All our funds go into our monthly equipment contracts and bandwidth. So if anyone out there works for Palm / Microsoft / Compaq / Dell / Danger / etc and wants to see Feedster support on your platform, we're happy to accommodate. Just drop us an email for a shipping address and we'll get going on it. All we need is a mobile device with a web browser, a detectable user agent and some type of wifi / Internet connectivity.
So many devices, so little time....
Wed, 30 Jul 2003 20:33:41 GMT
Heh. Hopefully you'll like these buttons as much as I do. And I know I need to add them to my own blog (note to self). Still feel free to add them to yours. Thanks to www.kalsey.com for the button maker UI and Paint Shop Pro for the rest.
Wed, 30 Jul 2003 17:39:30 GMT
With today's round of three blog entries, clearly I don't need any help from the peanut gallery* to tell me I'm dumb -- I'll do it myself -- and then retract it myself. The issue at hand is encoding binary files in syndicated content. I've recently learned that Atom allows for this -- and from a pure technical perspective, I'll admit its elegant. But from a practical standpoint, I think its rife with issues as I elaborated here. Well in the comments to that blog entry I saw a bunch of feedback and eased back on my position. The primary reason I did so was Dave pointing out to me that Enclosures are how RSS handles the issue and when I quickly (and poorly) read the enclosure spec, I walked away thinking that RSS did it the same way as Atom and then RSS would suffer from the same objection. And since that wasn't fair to Atom, I retracted.
Then it was pointed out to me that I was actually wrong in how enclosures work -- they actually work correctly -- pointers to content as opposed to content itself. And this brings me back to my original position.
So what took so long? The usual -- I got involved in debugging a really complex bit of string code and that sucked up the morning. Then lunch. Then I honest to god stubbed my toe (and almost ripped the nail off). And that brings us up to date. So I'm resuming my original position -- encoded content within an Atom feed is just opening the door to a whole bunch of issues. I understand why they're doing it but I don't agree.
I'd really like to hear an opinion from an aggregator vendor. Dave weighed in against it but what about Kevin, Nick, Greg and Luke? How do you guys feel? You're the ones who have to implement it. Are my fears valid?
*But heck feel free.
Wed, 30 Jul 2003 15:11:58 GMT
Sigh. I thought I was making a valid point about Atom (and I do still kinda think that) but Dave correctly pointed out that RSS supports enclosures too -- something that I didn't realize. And, thus, everything I said applies to RSS as well. Yeah I should have definitely known that -- considering what I do -- and I do know it -- but .... So this is an egregious example of overwhelming stupidity. I'm not proud. So consider this an apology. I shouldn't have labeled / titled the post as I did.
I'm tempted to just delete the previous post but that would lead to the whole arguments from a few weeks ago about post deleting / editing [Shudder].
Wed, 30 Jul 2003 13:34:30 GMTAtom: The Fatal Flaw No One Has Yet Noticed Now given that atom doesn't really exist yet, it may be early to make a draconian statement like "Fatal Flaw" but this is fairly awful from my perspective. I was being interviewed for Under the Iron and the interview brought this to my attention. The (most excellent) question was: Q: When Atom/Pie/nEcho is officially a spec and people start to use the real extensive metadata possible with it, will you need to adapt Feedster to searching more efficiently? Also, since Atom/Pie/nEcho supports content as sound files, pictures and even video, do you plan to look into those files and index their metadata as well? For the answer to the 1st part of the question, you'll have wait for the interview. But the flaw to me was here "Atom supports content". Now I don't have any problem with content being supported -- but -- Atom supports encoded content. And that, to me, is a fatal flaw. Here's the entry from the Atom wiki at: http://www.intertwingly.net/wiki/pie/content?action=highlight&value=encoding. And here's a sample entry:
Hello, weblog world! 2 < 4!
Wed, 30 Jul 2003 11:06:39 GMT
Jason has more on colors (I mentioned the matched color picker recently). Nice stuff.
Wed, 30 Jul 2003 11:01:48 GMT
I've been a user of Windows related multimedia files (originally .AVI) since like 1992ish -- back in the CD-ROM days and I just don't get it. Whenever I'm on the net and I encounter .MPEG or .MOV / .QTW / QuickTime video files they just plain work. Period. Rarely if ever does an MPEG file not play. Same for QuickTime. Why is it then that I rarely, if ever, can make a .AVI file just play? I'll see a .AVI file on the net and then I download it and nothing. I always see "Attempting Codec Download" and then "Error". And then, utterly annoyingly, Windows makes the less than intelligent decision "I'll just play the audio track". Sure that makes sense. After all when you can't show the pictures, just give them the sounds. Sigh.
Now I'll do what all good bloggers do whenever they enchounter a Windows issue and make the anguished cry of pain! Scoble ....!!!!
Note 1: Bitching was done in the context of a Fresh Windows XP Pro install and a new installation of Windows Media 9. I mean I did the damn update last night. Don't get much fresher than that. Piffle. When my clients ask me about streaming media, I guess my answer is here.
Note 2: Oh and why did I say "What if no one ever told you?" Well what kinds of video files are mostly on the net? Think anyone would really complain?
Tue, 29 Jul 2003 18:51:11 GMT
Sorry for being so quiet. I've been re-architecting / re-implementing our meta tag indexer and suffering through the hell that is metatags. We try and be flexible but man, oh freaking man. When I look at the cruft that is metatags I just want to wince, moan, scream, pound my big right toe with a mallet and then scream some more. Anyway . And don't even get me started on this annoying new thing where rssurls will sometimes have a ? after them and sometimes not. Hopefully more later.
Hey! Its Me! Thanks Chris!
Mon, 28 Jul 2003 15:17:24 GMT
There are a lot of interesting blogs out there --- all with rss feeds. And Feedster, with what is now I suspect the single largest database of RSS feeds available, sees so, so many of them. Francois and I have been wrestling with this and, at least for now, we've added to the Feedster home page a "Blog of the Day" link that we'll try and rotate daily (doh!). If you want to make suggestions then use interestingblog @ feedster.com
Thanks to Mike in Australia for doing the Blog if the Day graphic.
Mon, 28 Jul 2003 13:05:14 GMT
How do you describe a conference, Gnomedex, where you get to meet and hang out with:
Over the next few days I'll blog more about the conference including slides I thought were relevant. The overall impression I got from the conference was simple -- it rocked. Chris managed to make a technical conference **fun**. How often do you see that? And I have to admit that I was purely in awe of watching Chris' people skills. For me the most interesting part of the conference was probably just watching Chris interact with people. He's a natural -- its so clear why people pretty much love him.
So if you missed Gnomedex then all I have to say is you should make it next year. The event was outstanding. Thanks Chris.
Thu, 24 Jul 2003 14:34:05 GMT
Carrying out my silly (but functional) tradition of a different blogging tool for travel, my Gnomedex blog is here:
And apparently my Gnomdex Daily News is causing trouble. :-)
Thu, 24 Jul 2003 14:18:14 GMT
Well Gnomedex keeps drawing closer and getting ready to be away (again) and doign some special stuff for Gnomedex keeps knocking blogging off my list. Sorry about that. Here are some interesting links:
Thu, 24 Jul 2003 11:22:16 GMT
Check out http://www.feedster.com/news/ when you get a chance. What we've done here is dynamically (1/2 hour to 1 hour basis) create a "Blogpaper" a virtual newspaper assembled out of different blog entries all with a common theme. And, of course, the theme is Gnomedex. So if you're a blogger and talking about Gnomedex, this could well feature you.
Note: The dynamic nature of this won't really show up until people start posting when Gnomedex starts.
Tue, 22 Jul 2003 17:52:21 GMT
I've had some interesting IM chats with Pieter and he's updating GGSearch for Feedster. GGSearch is a client side (for Windows) tool for more easily building Google searches and its actually pretty darn cool. The way it works is it builds up a search query and then sends it to the browser. Pieter made a conscious decision to avoid the Soap api for Google and he's going to apply the *same approach to Feedster -- just build up a url. Give Pieter any comments you have on how a client side API for Feedster might work.
*Note: We do have a Soap API for Feedster although it is currently undocumented.
Tue, 22 Jul 2003 16:12:11 GMT
I just wanted to let people know that we've now partially implemented Robots.txt. Now before anyone flames on, I can just let you know that the main Feedster spider, the thing that checks sites regularly, is using Robots.txt. We're grabbing it once per week, every Sunday. Please note that there is NO standard for how frequently you grab it and this is what we decided. If more frequent is needed, let us know.
And please bear in mind that if you use robots.txt to turn off our indexing of your blog (which is fine) then we will still periodically check your url to make sure that you haven't changed your mind. If you want us to just go the heck away, never touch you, etc., then you have to let us know personally since we'll flip the database bit that says "A real live human made an intelligent reasoned decision to opt out of Feedster so we're never bothering them again unless they specifically ask us too".
As per the ill defined robots.txt spec we look for robots.txt in the root directory of a weblog. If your blog is located at http://radio.weblogs.com/0103807/ then this means that we're looking for the file http://weblogs.com/robots.txt. Subdomains and blogs are way, way, way too random for us to check every possible location.
Now where Robots.txt isn't doing yet is handling images. The reason for this is we use a separate crawler to handle Images and, in particular, we have to do a bunch of path analysis to compare relative to absolute paths and the whole issue of www and no www. So while we're making progress, the current implementation is, indeed, buggy and hence turned off. We'll get it in as soon as possible but with Gnomedex this week and my traveling to, of all places, Des Moines, Iowa, this isn't really likely. At least not a stable, reliable version.
Tue, 22 Jul 2003 15:40:25 GMT
I've had a flurry of requests recently asking why Feedster doesn't have a Technorati like "Cosmos". So over the past few days, with the assistance of Mike from down under who tossed in a critical string parser, I've been implementing this. And we're calling it your "Feedster Web". We looked at the whole Nebula / Galaxy / Constellation / etc but Richard Soderberg came up with "Web" and that one stuck. Now while the final code may not be ready until after Gnomedex (there's one bug that makes some domains non-"webbable"), I thought I'd at least give you a preview so you can tell me if we're on track.
Please leave comments on this blog entry and I'll see if we can, well, accommodate them.