Feeds per user pony! November 10, 2005 4:04 AM Subscribe
Um, a *meaningful* continuation, I mean.
posted by goodnewsfortheinsane at 4:07 AM on November 10, 2005
posted by goodnewsfortheinsane at 4:07 AM on November 10, 2005
Awesome. He shoulda just posted an RFP (Request for Pony) to MeTa and let the full-time Mathowie take car of it.. Let me now demonstrate how this can be abused. Example:
Plutor is so awesome. …
(I didn't use so you could see my awesome abuse method.)
posted by Plutor at 4:13 AM on November 10, 2005
Plutor is so awesome. …
(I didn't use so you could see my awesome abuse method.)
posted by Plutor at 4:13 AM on November 10, 2005
I don't use RSS, never had, I've no chuffing idea what this is all about, but sjvilla was kind enough to credit me with the idea of something invisible, so I'm all for it.
Well done, sir.
posted by NinjaPirate at 4:18 AM on November 10, 2005
Well done, sir.
posted by NinjaPirate at 4:18 AM on November 10, 2005
Okay, I get the mechanics of it now (although it was pretty obvious to begin with). But what's the point? Why would you want a per-user feed in the first place?
posted by chrismear at 4:52 AM on November 10, 2005
posted by chrismear at 4:52 AM on November 10, 2005
couldn't you do things like put on your web page a list of your most recent comments to mefi?
posted by andrew cooke at 4:56 AM on November 10, 2005
posted by andrew cooke at 4:56 AM on November 10, 2005
That would be neat. But this grabs posts, not comments, and then only from the last 24 hours of MeFi. Which is why I got the impression that it this is more of a notification thing, rather than an archive/log thing.
Anyway, I'm going to hold my tongue now and wait for sjvilla79 to explain.
posted by chrismear at 5:06 AM on November 10, 2005
Anyway, I'm going to hold my tongue now and wait for sjvilla79 to explain.
posted by chrismear at 5:06 AM on November 10, 2005
Is this something you'd need to be nude sleepwalking on Murelax to understand?
I can't imagine a more convoluted system for acheiving this.
Except maybe mathowie dispatching a daily carrier pigeon to the home of each user who requests updates.
While I don't know much about this sort of a thing, wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every so often and generate a feed that way?
posted by jack_mo at 5:21 AM on November 10, 2005
I can't imagine a more convoluted system for acheiving this.
Except maybe mathowie dispatching a daily carrier pigeon to the home of each user who requests updates.
While I don't know much about this sort of a thing, wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every so often and generate a feed that way?
posted by jack_mo at 5:21 AM on November 10, 2005
That's it? Where are the brass bands and righteous indignation. (I'm glad you're feeling better sj, but c'mon, give us something to work with.)
posted by OmieWise at 5:26 AM on November 10, 2005
posted by OmieWise at 5:26 AM on November 10, 2005
Except maybe mathowie dispatching a daily carrier pigeon to the home of each user who requests updates.
Mine comes by owl.
posted by terrapin at 5:30 AM on November 10, 2005
Mine comes by owl.
posted by terrapin at 5:30 AM on November 10, 2005
RFC 1149 - Standard for the transmission of IP datagrams on avian carriers.
posted by loquacious at 5:41 AM on November 10, 2005
posted by loquacious at 5:41 AM on November 10, 2005
On second reading of sjvilla79's lengthy treatise, the answer to my question above would appear to be 'Yes'.
If those search_posts pages had divs for each post and an id or something on the 'Go to the detail view for this result', even a total codemong like me could sort out an XSLT stylesheet to transform the page into RSS, so you could just grab the page once a day and whack it on your server, whizz it through the XSLT, and look at it in your aggregator. No one would need to opt-in by adding code to posts this way, either (which is the really daft bit of the proposed plan, not to mention abuse - 'Ooh, I think I'll be RSS:User_ID=1 today!').
(Please note: I don't really know what I'm talking about, but that would seem do-able, no?)
On preview:
Mine comes by owl
Bloody low user number folk and their special treatment ;-)
posted by jack_mo at 5:42 AM on November 10, 2005
If those search_posts pages had divs for each post and an id or something on the 'Go to the detail view for this result', even a total codemong like me could sort out an XSLT stylesheet to transform the page into RSS, so you could just grab the page once a day and whack it on your server, whizz it through the XSLT, and look at it in your aggregator. No one would need to opt-in by adding code to posts this way, either (which is the really daft bit of the proposed plan, not to mention abuse - 'Ooh, I think I'll be RSS:User_ID=1 today!').
(Please note: I don't really know what I'm talking about, but that would seem do-able, no?)
On preview:
Mine comes by owl
Bloody low user number folk and their special treatment ;-)
posted by jack_mo at 5:42 AM on November 10, 2005
Oop, do I have XSLT ass-backwards there? A Googling suggests so. Sorry!
posted by jack_mo at 5:53 AM on November 10, 2005
posted by jack_mo at 5:53 AM on November 10, 2005
I'd rather provide per-user RSS feeds myself than have everyone stuff pointless HTML into everything they post. It'd save resources since no one would have to scrape the site to create these feeds.
posted by mathowie (staff) at 6:11 AM on November 10, 2005
posted by mathowie (staff) at 6:11 AM on November 10, 2005
#: Is this something you'd need to be nude sleepwalking on Murelax to understand?
No, I don't think so. It's a simple concept and something that regular RSS feed readers (the people kind) should immediately see beneficial. <joking>I have heaps of pills if anyone wants some though. $5 a tab for Valium or Murelax. Take your pick.</joking>
#: wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every....
Um, no it wouldn't. Scraping is bad. Matt would definitely be pissed with hundreds of MeFi members scraping their threads. Plus, a dude has already done this. See MetaRSS.
#: That's it? Where are the brass bands and righteous indignation.
It would have been better had the Atom thing not popped up, I guess.
#: I got the impression that it this is more of a notification thing, rather than an archive/log thing.
That depends on how often MeFi's feeds update and how often your filtering service chooses to poll the feeds. I've noticed a search only takes about five minutes to appear in FeedShake after posting via MeFi's main feed. And yeah you could essentially do whatever you want with the syndicated content (send via RSS to email, archive them, post them on a blog's link-roll and more). Lastly, the posts are in full and filtered just the way you want them. That's why I think this is cool. Well, uber cool.
#: Let me now demonstrate how this can be abused.
I assumed better of MeFi's members than that. This isn't fucking MySpace. Anyway, surely abuse would mean same kind of reaction for abusing the normal rules though. Straws. Clutching.
I also don't see how this is abuse anyway. I mean you're only fooling yourself by adding the code in spite of the system. For example, who would be subscribing to your feed anyway? You need to post good stuff if people are going to syndicate you. You'd obviously know that trolls are often ignored. Teased and pervertedly poked and prodded, sure. But, um, ultimately ignored.
#: I've no chuffing idea what this is all about, but sjvilla was kind enough to credit me with the idea of something invisible, so I'm all for it.
No problem. New to RSS and feed reading? Start via Google.
It has just past 1 in the morning here and I'm going to bed soon. Yes, the irony.
posted by sjvilla79 at 6:15 AM on November 10, 2005
No, I don't think so. It's a simple concept and something that regular RSS feed readers (the people kind) should immediately see beneficial. <joking>I have heaps of pills if anyone wants some though. $5 a tab for Valium or Murelax. Take your pick.</joking>
#: wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every....
Um, no it wouldn't. Scraping is bad. Matt would definitely be pissed with hundreds of MeFi members scraping their threads. Plus, a dude has already done this. See MetaRSS.
#: That's it? Where are the brass bands and righteous indignation.
It would have been better had the Atom thing not popped up, I guess.
#: I got the impression that it this is more of a notification thing, rather than an archive/log thing.
That depends on how often MeFi's feeds update and how often your filtering service chooses to poll the feeds. I've noticed a search only takes about five minutes to appear in FeedShake after posting via MeFi's main feed. And yeah you could essentially do whatever you want with the syndicated content (send via RSS to email, archive them, post them on a blog's link-roll and more). Lastly, the posts are in full and filtered just the way you want them. That's why I think this is cool. Well, uber cool.
#: Let me now demonstrate how this can be abused.
I assumed better of MeFi's members than that. This isn't fucking MySpace. Anyway, surely abuse would mean same kind of reaction for abusing the normal rules though. Straws. Clutching.
I also don't see how this is abuse anyway. I mean you're only fooling yourself by adding the code in spite of the system. For example, who would be subscribing to your feed anyway? You need to post good stuff if people are going to syndicate you. You'd obviously know that trolls are often ignored. Teased and pervertedly poked and prodded, sure. But, um, ultimately ignored.
#: I've no chuffing idea what this is all about, but sjvilla was kind enough to credit me with the idea of something invisible, so I'm all for it.
No problem. New to RSS and feed reading? Start via Google.
It has just past 1 in the morning here and I'm going to bed soon. Yes, the irony.
posted by sjvilla79 at 6:15 AM on November 10, 2005
Might I be the first to point out that the underscore character, the colon, and the equals sign are invalid characters for an ID attribute, according to the CSS specification? (IIRC, they're fine by HTML standards, but a strict CSS parser will hate them. And IIRC again, I think that underscore may be in the process of being amended into CSS2, but the rest aren't.)
posted by delfuego at 6:22 AM on November 10, 2005
posted by delfuego at 6:22 AM on November 10, 2005
sjvilla79, your post ends in a link that goes nowhere for HTML users. That kind of kills it IMO.
posted by mathowie (staff) at 6:23 AM on November 10, 2005
posted by mathowie (staff) at 6:23 AM on November 10, 2005
#: Ooh, I think I'll be RSS:User_ID=1 today!
That could be an issue. You get to choose your own ID though. The RSS:User_ID=X thing was just a suggestion. I did say this in my main post on this topic (e.g. I was hoping the MeFi community could help improve the idea). If you can do better, well, go for it dude. Anyway, this is a simple hack for simple people. It would work well on a small scale, for example, and tools need not apply.
posted by sjvilla79 at 6:23 AM on November 10, 2005
That could be an issue. You get to choose your own ID though. The RSS:User_ID=X thing was just a suggestion. I did say this in my main post on this topic (e.g. I was hoping the MeFi community could help improve the idea). If you can do better, well, go for it dude. Anyway, this is a simple hack for simple people. It would work well on a small scale, for example, and tools need not apply.
posted by sjvilla79 at 6:23 AM on November 10, 2005
This whole thing is terrifying, like the end of 2001. Can I go back to bed now?
posted by selfnoise at 6:28 AM on November 10, 2005
posted by selfnoise at 6:28 AM on November 10, 2005
#: sjvilla79, your post ends in a link that goes nowhere for HTML users. That kind of kills it IMO.
But there's two versions I'm proposing, Matt. First there's the hidden code version and second there's the version with the ellipsis. Anyway, what's wrong with hiding the anchor tag at the end of your post when posting? Nobody will ever see it. Only the filtering service will pick it up and pluck whatever posts you've made from MeFi's main feeds.
posted by sjvilla79 at 6:31 AM on November 10, 2005
But there's two versions I'm proposing, Matt. First there's the hidden code version and second there's the version with the ellipsis. Anyway, what's wrong with hiding the anchor tag at the end of your post when posting? Nobody will ever see it. Only the filtering service will pick it up and pluck whatever posts you've made from MeFi's main feeds.
posted by sjvilla79 at 6:31 AM on November 10, 2005
If we can all code MeFi into our vision, I want more cowbell.
posted by tommasz at 6:46 AM on November 10, 2005
posted by tommasz at 6:46 AM on November 10, 2005
#: wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every....Why is scraping bad? jack_mo isn't suggesting scraping threads, which is apparently what MetaRSS does. Rather, he's providing a (IMO) better solution for providing per-user feeds, by scraping the search_posts or search_comments pages. That gives you a list of the most recent posts & comments by a particular user. That's what you want right?
Um, no it wouldn't. Scraping is bad. Matt would definitely be pissed with hundreds of MeFi members scraping their threads. Plus, a dude has already done this. See MetaRSS.
Relying on proper user behaviour is bad. Not to mention that
id attributes must be unique in a document, so even allowing users to write a post containing id attributes is bad.On preview, odinsdream.
posted by ijoshua at 6:52 AM on November 10, 2005
Allowing users to write their own unique id seems like a coffin nail to me, too. Never trust users.
posted by PantsOfSCIENCE at 7:02 AM on November 10, 2005
posted by PantsOfSCIENCE at 7:02 AM on November 10, 2005
Matt, if you're serious about providing per-user feeds, or if you want to pick back up on per-thread feeds, I'd be happy to offer my programming skillz. I do PHP & MySQL full-time, but I've seen some ColdFision. Contact me by email if you're interested.
posted by ijoshua at 7:04 AM on November 10, 2005
posted by ijoshua at 7:04 AM on November 10, 2005
I can really dig this idea if it's built into the posting system itself. Relying on users to paste their own code seems like a bad idea (laziness, abuse, ill-formed code).
posted by goodnewsfortheinsane at 7:12 AM on November 10, 2005
posted by goodnewsfortheinsane at 7:12 AM on November 10, 2005
#: wouldn't it be a hell of a lot easier to scrape the http://www.metafilter.com/search_posts.cfm?user_ID=xxxxxx type pages every....For the sake of clarity, MetaRSS does not use search_posts.cfm. The system is coded in such a manner that is friendly to the servers, even if hundreds of people were to start using it. For every AskMe or MeFi thread it is told to monitor, MetaRSS only makes one update, per hour, for all users who subscribe to that feed.
Um, no it wouldn't. Scraping is bad. Matt would definitely be pissed with hundreds of MeFi members scraping their threads. Plus, a dude has already done this. See MetaRSS.
In case it's not clear, MetaRSS does not result in per user feeds. Instead, it allows anyone to create an RSS feed of any AskMe or MeFi thread so that they can keep up to date with it via RSS instead of refreshing until new posts appear. It does not monitor all threads, only threads that users request.
I also provide an RSS feed for MeTa, but have not created a thread monitoring service for MeTa.
If there is a request for per user feeds, I'm sure I can whip something up, but my hesitation has been that, unlike threads, users don't stop posting after thirty days. I suppose I could limit the number of updates made per day to cut down on bandwidth, but it seems awfully consumptive to hit search_posts.cfm every hour or even day given that not everyone posts that frequently.
posted by sequential at 7:18 AM on November 10, 2005
Can we just be patient and wait for Matt to do this stuff right instead of OMG EL3TE HAZZORS BONERZ?
posted by selfnoise at 7:20 AM on November 10, 2005
posted by selfnoise at 7:20 AM on November 10, 2005
The fact that my post has roused some action in the MetaFilter RSS department is satisfying enough for me. I'll use my method until something better comes along, though. And I guess some other members might choose to use it too. Night.
posted by sjvilla79 at 7:20 AM on November 10, 2005
posted by sjvilla79 at 7:20 AM on November 10, 2005
sequential, didn't mean to muddy the waters on MetaRSS.
posted by sjvilla79 at 7:23 AM on November 10, 2005
posted by sjvilla79 at 7:23 AM on November 10, 2005
Can we just be patient and wait for Matt to do this stuff right instead of OMG EL3TE HAZZORS BONERZ?I don't know if this was in response to me, but MetaRSS was created in March of this year. At that time, I'd have to have had faith, not patience, that Matt would ever address the particular need I had.
sequential, didn't mean to muddy the waters on MetaRSSNo offense taken, sjvilla79. I just wanted to be clear that MetaRSS doesn't do anything nasty. In fact, it does exactly what was requested in this thread.
I'd rather provide per-user RSS feeds myself than have everyone stuff pointless HTML into everything they post. It'd save resources since no one would have to scrape the site to create these feeds.Though I would prefer such a system to come from the Metafilter server, I could provide such a service through MetaRSS, probably even today, without your servers or bandwidth.
posted by sequential at 7:38 AM on November 10, 2005
without your servers or bandwidth.That was supposed to say "without taxing your servers or bandwidth."
posted by sequential at 7:51 AM on November 10, 2005
Scraping is bad
True, but your idea would mean extra requests for the RSS feeds, which comes to the same thing. And it could even put more of a strain on the server that the search_posts way: Each time you request your feed from [Feedshake's] server, subfeeds are parsed,filtered and merged into a unique RSS 2.0 feeds.
And using those search_posts pages is still about a gazillion times easier than jumping through hoops adding code to every post, setting up feedshake etc. (which doesn't appear to actually work either - your one specially-tagged post so far doesn't show up in the Feedshake feed linked at the bottom of your explanation page.)
You get to choose your own ID though. The RSS:User_ID=X thing was just a suggestion.
Then how would anyone know how to subscribe to your feed?
I have heaps of pills if anyone wants some though. $5 a tab for Valium or Murelax. Take your pick
I'm just out of hospital after my near-annual lung collapse horrorscare and have some totally wild opiates - wash down with a Red Stripe, and you're right off it for a good few hours. Swopsies? ;-)
posted by jack_mo at 8:23 AM on November 10, 2005
True, but your idea would mean extra requests for the RSS feeds, which comes to the same thing. And it could even put more of a strain on the server that the search_posts way: Each time you request your feed from [Feedshake's] server, subfeeds are parsed,filtered and merged into a unique RSS 2.0 feeds.
And using those search_posts pages is still about a gazillion times easier than jumping through hoops adding code to every post, setting up feedshake etc. (which doesn't appear to actually work either - your one specially-tagged post so far doesn't show up in the Feedshake feed linked at the bottom of your explanation page.)
You get to choose your own ID though. The RSS:User_ID=X thing was just a suggestion.
Then how would anyone know how to subscribe to your feed?
I have heaps of pills if anyone wants some though. $5 a tab for Valium or Murelax. Take your pick
I'm just out of hospital after my near-annual lung collapse horrorscare and have some totally wild opiates - wash down with a Red Stripe, and you're right off it for a good few hours. Swopsies? ;-)
posted by jack_mo at 8:23 AM on November 10, 2005
Well, I asked for per-thread RSS templates, but there wasn't really a consensus on what met with RSS standards and worked in popular readers. I could offer a RSS feed for every thread today if I had a good template to stuff the content into.
posted by mathowie (staff) at 9:05 AM on November 10, 2005
posted by mathowie (staff) at 9:05 AM on November 10, 2005
Thanks for the reply, mathowie. If my comments came across as snarky, please note that absolutely no slight was intended. In fact, the reason I created MetaRSS and the reason I offered to continue its development is because I understand your time is valuable, to me and to the community. It's not my style to ask others to do what I can do for myself.
Perhaps I still misunderstand the issue that caused the lack of consensus. I've done a significant amount of QA on the feeds, using Bloglines, Yahoo, NewsGator, Firefox (live bookmarks), and Thunderbird, which are, according to my professional experience offering RSS feeds, the most popular RSS readers. (Firefox and Thunderbird are not very popular, they are just a part of the development suite of software to test RSS on and what I use personally.)
I'd be happy to test other RSS readers and make the necessary modifications to the XML to make it work. Any suggestions on what RSS readers MeFi users prefer?
Here are some example feeds produced by MetaRSS: If you need an Atom version, I could create one as well.
Please note that those are links to a development version of the code. I am, as I write, working on fixing various bugs, like relative URLS, which happen to cause the production feeds to not validate. This version of the code will disappear some time in the next few days. The production version of the code is available minus the "-dev" in the filename.
posted by sequential at 9:48 AM on November 10, 2005
Perhaps I still misunderstand the issue that caused the lack of consensus. I've done a significant amount of QA on the feeds, using Bloglines, Yahoo, NewsGator, Firefox (live bookmarks), and Thunderbird, which are, according to my professional experience offering RSS feeds, the most popular RSS readers. (Firefox and Thunderbird are not very popular, they are just a part of the development suite of software to test RSS on and what I use personally.)
I'd be happy to test other RSS readers and make the necessary modifications to the XML to make it work. Any suggestions on what RSS readers MeFi users prefer?
Here are some example feeds produced by MetaRSS: If you need an Atom version, I could create one as well.
Please note that those are links to a development version of the code. I am, as I write, working on fixing various bugs, like relative URLS, which happen to cause the production feeds to not validate. This version of the code will disappear some time in the next few days. The production version of the code is available minus the "-dev" in the filename.
posted by sequential at 9:48 AM on November 10, 2005
>>Let me now demonstrate how this can be abused.
>I assumed better of MeFi's members than that.
You're new around here, aren't you.
Wait until someone figures out how to put the goatse.xc man in their neighbour's feed. (Or has that already happened?)
posted by timeistight at 11:01 AM on November 10, 2005
>I assumed better of MeFi's members than that.
You're new around here, aren't you.
Wait until someone figures out how to put the goatse.xc man in their neighbour's feed. (Or has that already happened?)
posted by timeistight at 11:01 AM on November 10, 2005
Well, I've had my own (sort-of) take on this with MefiBookmarks.
posted by FieldingGoodney at 11:50 AM on November 10, 2005
posted by FieldingGoodney at 11:50 AM on November 10, 2005
Per-user RSS? Oh great, something else I can be unpopular at!
posted by clevershark at 4:38 PM on November 10, 2005
posted by clevershark at 4:38 PM on November 10, 2005
Any suggestions on what RSS readers MeFi users prefer?
I imagine there's a fair few of us using NetNewsWire?
Your AskMe feeds look perfect in it, but in both MeFi feeds, the text "http://www.metafilter.com" shows up at the beginning of the description of the post (as opposed to the following comments), even though it's not there in the description element when I view source. (Little screenshot in case that doesn't make sense.)
Per-thread feeds would be lovely, much lovelier than per-user...
posted by jack_mo at 5:39 PM on November 10, 2005
I imagine there's a fair few of us using NetNewsWire?
Your AskMe feeds look perfect in it, but in both MeFi feeds, the text "http://www.metafilter.com" shows up at the beginning of the description of the post (as opposed to the following comments), even though it's not there in the description element when I view source. (Little screenshot in case that doesn't make sense.)
Per-thread feeds would be lovely, much lovelier than per-user...
posted by jack_mo at 5:39 PM on November 10, 2005
jack_mo, I just downloaded a demo of NetNewsWire, installed it on my iBook, and loaded up the RSS feed in question. Unfortunately, I can not confirm the same behavior, but I believe that is because I've corrected the problem since the time you subscribed to the feed and the time that I tried to confirm the bug.
Please try subscribing to another feed and let me know if you experience the same behavior. (Feel free to contact me via email.) I'll happily continue to look into this further if this continues.
posted by sequential at 6:47 PM on November 10, 2005
Please try subscribing to another feed and let me know if you experience the same behavior. (Feel free to contact me via email.) I'll happily continue to look into this further if this continues.
posted by sequential at 6:47 PM on November 10, 2005
How about an RSS/Atom feed for the my comments (and maybe myposts) page. That would be nice, but I guess authentication issues might come up.
posted by gsb at 3:28 AM on November 11, 2005
posted by gsb at 3:28 AM on November 11, 2005
#: I'm just out of hospital after my near-annual lung collapse horrorscare and have some totally wild opiates - wash down with a Red Stripe, and you're right off it for a good few hours. Swopsies? ;-)
Yes, that'd be nice. How to mail these things though.
posted by sjvilla79 at 7:07 PM on November 14, 2005
Yes, that'd be nice. How to mail these things though.
posted by sjvilla79 at 7:07 PM on November 14, 2005
You are not logged in, either login or create an account to post comments
Haven't tried it (it seems to involve adding some HTML to posts), but it's obviously a continuation of the closed thread below, so I thought I'd post it.
posted by goodnewsfortheinsane at 4:05 AM on November 10, 2005