Last modified: 2013-01-23 13:40:35 UTC
Certain templates such as the four main parts of the main wikipedia.org page (featured article, on this day, news, did you know) should be able to be fed as RSS. Rather, the software should update XML pages when the templates are changed. I have set severity to normal because I believe that this is a very important feature. The added "bugs" relate to this.
*** Bug 7746 has been marked as a duplicate of this bug. ***
*** Bug 4372 has been marked as a duplicate of this bug. ***
*** Bug 4818 has been marked as a duplicate of this bug. ***
Actually even if the related changes page ([[Special:Recentchangeslinked]]) had an RSS feed, like [[special:recentchanges]], it would be useful. (It'd be even more useful if there was an rss feed for related changes, but only counted category additions, like in dpl's with order mode set to categoryadd)
That's definatley another template that would see some good RSS use. And I like the idea of being able to select which items you want to be fed (in case you don't want to have 30 wiki-feeds), however I'm not sure it would be completely appropriate to have every item in the same feed; i.e, perhaps article-content feeds should be seperated from article-edit feeds. Thoughts? To get some discussion going on this topic, here are a few things to think about... Any thoughts on whether or not every page should have a feed or just specific types (templates), or how the wiki-admin can distinguish which templates they would like to have rss-feedable? I'm not sure if there is a template editor/creator in the mediawiki package, but if there is, could this be a good place to have this feed option? Also, what other common templates throughout the mediawiki package would benefit from having a feed?
Warning: there is a fair probability that I don't know what I'm talking about. What do you mean by template?, I was under the impression that a template (in mediawiki) was any page transcluded like {{so}}. (usually in the template: namespace). They are just ordinary pages. Therefor, I don't see how you can make an RSS feed of it, with the exception of being a feed of changes (diffs) to that page (which [http://en.wikinews.org/w/index.php?title=Main_Page&action=history&feed=rss already exists]). Unless you do something like I suggested at [[bugzilla:8490#c2]] (which I also have no idea how feasible such a thing would be, or weather its a good or bad idea). I just don't see how you could get useful feeds of any random page in a wiki just based on their ordinary contents. Special pages on the other hand would probably (I assume) be much more easy to create rss feeds for, as they're structured content, not just a random assembeledge of words. -[[n:user:Bawolff]]
Perhaps template isn't the correct term. Let me try to clarify my thoughts. For actual feeds, "new material" is what I meant, instead of the fact that "material has changed" (though that could be useful) and what the change was. Because of this, there may need to be a way to distinguish an 'entry area' from an 'edit area'. Point being, supposing you can tell that it is an entry, the entry could have a feed. Now, instead of template, I mean to say specific pages (or specific places) that were clearly designed differently than the standard article. This was probably done because the nature of the content is quite different, mostly being temporary-entry data. I don't know if this is what you refer to as a Special page or not, but more structured content would obviously be of focus. Even specified tables within the page rather than the whole page itself. Examples: "featured article", "on this day", "news", "did you know", "current events" and it's subsidiaries, "featured content" and it's subsidiaries, updated entries in "portals", "wikiproject" updates and posts, the "recent changes" you mentioned, etc etc etc, the list can go on. Keep in mind, you wouldn't want to feed the whole page, only the pertinent entry. If so, the idea I have would be to be able to have an RSS option somewhere so that when an appropriate area is created, users can grab that entry in XML. Perhaps a tag would do, something like {{Feedable Data ID=1 Title="Title"}} "article text" {{End-Feedable Data}}, representing feed #1 on that page. Somewhere nearby you would have the RSS icon or link, and when it is pressed, it generates a page with appropriate XML. Something like /Feed:Generate&Featuredcontent%1 At first hand, since these data entries are so small, it doesn't look like a very useful tool; they look more like feed-snippets than feeds. But this is where things can get really interesting. Suppose you want to be able to grab a few snippets. Using some king of dynamic, multi-page parsing, XML generating script, someone could call an address that looked like .../Feed:GenerateMyFeed&feedId=Featuredcontent%1,news%4,sports%12 and it would generate a XML page that combined the feed-snippets you want to read into 'one big feed' with appropriate headers for each snippet, links, etc. There could even be a page that helped you put your numbers together and the order you want them in by creating the address through a formula. If you wanted your own custom feed through your account, you might have to login, your snippets and order would have to be saved in a database, other disadvantages, etc. But the great part about doing it this other way is that nothing ever has to be saved on the wiki, and any user could add any additional info they wanted provided its a variable that could be edited. Example: /Feed:GenerateMyFeed&title=Mikes_Feed&include=date,time&feedId=Portal:Radio%1 Reference this page, http://en.wikipedia.org/wiki/Portal:Radio Supposing that Portal:Radio had a tag like {{Feedable Data ID=1 title="Featured Article"}} for it's featured article, the above address might generate: Mikes_Feed Saturday, February 3rd, 3:45am PST Portal:Radio Featured Article "yadda yadda yadda yadda...Read more" I hope this got a few of my ideas out there in a more concise manner. But I'm also just spitting ideas out there.
Brian, you mentioned "supporting a good mechanism for building RSS/Atom feeds from content under user control", and I'd like to get your ideas on some of the above. What do you think, would something like this be feasible?
Just wanted to mention something along those lines would probably be useful to wikinews project. Be really cool if it worked with DPL's too (or even if we had a special:dplfeed?category=blah¬category=disputed&namespace=main etc (The arguments are what would normally be in a <DynamicPageList>). Personally how I would see something like that is if you have a normal page. Inside it you have some sort of feed extension <feed> title=foo content=bar ... </feed> It displays anything in content section of <feed> normally, as if its part of the rest of the page (or maybe that's toggleble), but if you goto w/index.php?title=baz&feed=rss (or perhaps something like special:feed/baz ) it turns it into an rss feed, based on the stuff inside the <feed></feed> tags.
Yes! Thats exactly what I mean. It would appear exactly how it should look on the page, but you can go to a specific address and just see that tagged area in XML. Would you be able to use those html-like tags <-->, or would you need the wiki-type {{--}}? Also, if you have two different places on a page you would like to be able to add feed tags to, they would have to be numbered somehow, first feed, second feed, etc. {{feed id=1 title="foo"}} ... ... ... {{end feed}} Then use this to isolate an area in a feedable XML document /special:feed=baz%1 /special:feed=baz%2 Or even use this to have both areas generated in a single feed special:generateMyFeed=baz%1,baz%2
Note: I am not a developer nor do I know what I'm talking about in my previous post. I just noticed that for extensions it seems like xml style tags are used <> (like in [[meta:DynamicPageList]], [[meta:Help:Inputbox]], [[meta:EasyTimeline]] etc) and {{}} is only used for transclusions, variables, and parser functions.
Correct, but functions that the wikipedia page must parse in server-side programming won't be able to be done with html. HTML is parsed one time, in the browser. Sure you can write a script to analyze html, but why do that when the page is written in a server-side language with a database? That, and you can't make up html tags as you see fit, they are standardized. Much like how wikipedia knows that [[this]] really means find the link to the "this" article, [[some:function]] can be made-up to do whatever you want. At least, thats how I understand it. Perhaps it will need to be done with [[these]] square brackets.
(In reply to comment #12) The user was referring to the fact that most of our parser hook extensions appear to use XML-like tags, e.g. <ref></ref>, <references /> etc. This is because that is indeed how parser hook extensions are exposed to the users. Wiki markup is not treated like HTML, and goes through a lot of processing to turn it into HTML; we can introduce whatever new tags we like using this mechanism. Please can *all* further discussion on this bug be limited to those developing a solution, or those who have some comment pertaining to the implementation. That means no "me too" endorsements.
This bug would be solved with https://bugzilla.wikimedia.org/show_bug.cgi?id=20818
Note wikinews uses the gnsm extension for rss feed. See special:newsfeed?feed=atom on enwikinews. it can take parameters in a similar manner to dpl
(In reply to comment #0) > Certain templates such as the four main parts of the main wikipedia.org page > (featured article, on this day, news, did you know) should be able to be fed > as > RSS. Rather, the software should update XML pages when the templates are > changed. > I believe wikipedia has an extension installed to do this (featureFeeds or something like thst). This bug could perhaps be closed.
