Difference between revisions of "explain xkcd:Community portal/Coordination"

Explain xkcd: It's 'cause you're dumb.
Jump to: navigation, search
(Page names)
(To do list)
Line 53: Line 53:
 
** [[Special:DoubleRedirects]]
 
** [[Special:DoubleRedirects]]
 
**: (Took a chunk out of these the good ol' fashioned way, but there's got to be a wiff of Perl or Python to automate this... ? -- [[User:IronyChef|IronyChef]] ([[User talk:IronyChef|talk]]) 14:44, 9 August 2012 (UTC))
 
**: (Took a chunk out of these the good ol' fashioned way, but there's got to be a wiff of Perl or Python to automate this... ? -- [[User:IronyChef|IronyChef]] ([[User talk:IronyChef|talk]]) 14:44, 9 August 2012 (UTC))
 +
**:: Well, there's [https://sourceforge.net/apps/mediawiki/mwclient/ mwclient], a Python interface to the mediawiki API which I've used to move the comics to the new names. We could certainly create scripts to perform maintenance tasks and share the snippets here on the wiki. Automated tools will be useful while we establish standards early on. If you'd like help getting started, let me know. --[[User:Waldir|Waldir]] ([[User talk:Waldir|talk]]) 19:40, 9 August 2012 (UTC)
 
** convert [[Special:LinkSearch/en.wikipedia.org|wikipedia links]] to the <nowiki>{{w|Lorem ipsum}}</nowiki> format
 
** convert [[Special:LinkSearch/en.wikipedia.org|wikipedia links]] to the <nowiki>{{w|Lorem ipsum}}</nowiki> format
 
** use lowercase xkcd everywhere on the wiki (see [http://xkcd.com/about/ How do I write "xkcd"?])
 
** use lowercase xkcd everywhere on the wiki (see [http://xkcd.com/about/ How do I write "xkcd"?])

Revision as of 19:40, 9 August 2012

Community Portal
Dialog-information on.svg

Proposals (+post)
Place for ideas and suggestions to improve the wiki's design and organization on general issues.

Preferences-system.svg

Technical (+post)
Technical issues regarding the site, including bug reports or MediaWiki extensions requests.

Edit-find-replace.svg

Coordination (+post)
Community-managed page for coordinating content editing and maintenance tasks.

Tools-hammer.svg

Admin requests (+post)
Problems requiring assistance from an admin. User problems, changes to protected pages, etc.

Help-browser.svg

Miscellaneous (+post)
Place for general chit-chat about virtually anything that doesn't fit anywhere else.

View all community portal sections at once here
Hyperlink-internet-search.svg

Issue dates

Hi Jeff,

As i'm creating pages I struggle with the issue dates of comics. I've added a comment to all pages that contain the (unknown/incorrect) dates. Is there a way to research those dates? --Rikthoff (talk)

[1] if you mouse over the comic name, it will have the date. --Jeff (talk) 18:26, 3 August 2012 (EDT)

- if you mouse over comic name in "Archive" section of xkcd.com. Older comics(1-44 or so) might be found in livejournal archiveB. P. (talk) 18:35, 3 August 2012 (EDT)

Should we consider using "2012-08-03" style dates and letting localization "do the right thing"? Most pages so far use "August 3, 2012" style dates, with a few incorrectly doing "August 3rd, 2012"... Presumably the template could do the localizing/localising...--B. P. (talk) 18:39, 3 August 2012 (EDT)

The date is also available with the JSON API, which I'm going to use for the import. I use {{#dateformat: year-month-day}}, MediaWiki should figure out the correct way to display it based on your preferences. --SlashMe (talk) 18:47, 3 August 2012 (EDT)

Moved from User talk:Jeff. --Philosopher Let us reason together. 00:15, 4 August 2012 (EDT)

Date?

How do I find the date a comic was first posted (to put in the comic header here?) TheHYPO (talk) 12:26, 3 August 2012 (EDT)

Moved from Talk:Main Page. --Philosopher Let us reason together. 00:43, 4 August 2012 (EDT)

Original posting date is listed on xkcd's [archive page] as hover-text for each post. The first 44 comics are all listed as 2006-01-01. Many of these were previously posted on the [livejournal site], and some dates can be found/inferred by checking there.--B. P. (talk) 17:49, 7 August 2012 (UTC)

To do list

I suggest a todo list to be added here so newcomers will have an idea of concrete things they can do to help. I'll start by moving some items I've been collecting on my user page. Feel free to add more :)

Things to do

  • Complete all entries from the List of all comics
  • Special:WantedPages lists pages that have links to them but haven't been created yet.
  • More topics that could be covered here besides the comics themselves:
    • our twitter account
    • the xkcd irc channel (and its wiki)
    • the xkcd blag
    • the xkcd forum
    • other sites explaining xkcd ([2], [3], [4], [5], maybe invite members+content of the other wikis in once we're established?)

Maintenance

  • images:
    • All image filenames should to be moved to the capitalized versions. This will allow comics to automatically include the image based on page title.
    • Redirects should be made from the "number.png" format to the titled format.
    • Images should be categorized (comic images, character images, illustrations for the explanations, etc.)
  • categorization (make sure these lists are empty):
  • building the web of links:
  • other
    • Special:DoubleRedirects
      (Took a chunk out of these the good ol' fashioned way, but there's got to be a wiff of Perl or Python to automate this... ? -- IronyChef (talk) 14:44, 9 August 2012 (UTC))
      Well, there's mwclient, a Python interface to the mediawiki API which I've used to move the comics to the new names. We could certainly create scripts to perform maintenance tasks and share the snippets here on the wiki. Automated tools will be useful while we establish standards early on. If you'd like help getting started, let me know. --Waldir (talk) 19:40, 9 August 2012 (UTC)
    • convert wikipedia links to the {{w|Lorem ipsum}} format
    • use lowercase xkcd everywhere on the wiki (see How do I write "xkcd"?)

There are more maintenance reports at Special:SpecialPages, for inspiration :) --Waldir (talk) 06:45, 6 August 2012 (EDT)

Date categories

I'm not sure the "Comics by month", by weekday, etc. Will be much useful, unless for those interested in running some stats. It might be more interesting to have specific months, such as Category:Comics from May 2011 and so on. What do you think? --Waldir (talk) 06:45, 6 August 2012 (EDT)

That was actually next for me: #time:year-month, but I wanted to study the globalization implications. I prefer over-categorizing rather than under-categorizing, since it's comparatively cheap. The assumption is that categories are the same as tags on the old site, and that mediawiki affords us some extra ways to automatically categorize pages in addition to the manual forms starting to emerge (by character, by subject, etc.) To paraphrase an old prof: you can't study what you don't measure; I've been wanting to see if, for example, Monday comics deal certain subjects, while Friday comics deal with another, etc. Not everybody's cup of tea, but of value perhaps to some, and insanely cheap to support both mentally and for the software. -- IronyChef (talk) 13:51, 7 August 2012 (UTC)

Page names

I think we should use the comic number and the title as the page name. Like so: "112: Baring My Heart". This would allow comics to be sorted by order in categories, but the pages would still have human-readable names for those of us who don't memorize all xkcd comic numbers ;) Thoughts? --Waldir (talk) 07:23, 6 August 2012 (EDT)

I agree, for another reason: for instance YouTube could be either the title of a page explaining how YouTube is referenced in xkcd, or the title of the explanation for comic #202 (titled "YouTube"). I don't know if I'm being clear here, but as we do not control the titles of the comics, that could create confusion with other pages. So using something like 202: YouTube would ensure disambiguation without being really complicated or awkward... And actually prefixing the comic title with its number seems quite relevant to me.
Additionally, that would solve potential problems such as Exoplanets: comic 786 or 1071?
Cos (talk) 14:33, 6 August 2012 (UTC)
Beat me to the punch; agreed. Numbers are unique and sequential, but not altogether that meaningful. Names are meaningful but (as we've seen) not unique. Some combination of both would be called for. We'd need to have the plain numbers redirect to the new topic (some double-redirects would need to be fixed up?) and the names would too (with at least one disambiguation page for now, and who knows: maybe more to come?) -- IronyChef (talk) 13:55, 7 August 2012 (UTC)
Following up on the YouTube discussion above, I'm wondering if we should leverage namespaces more: main:topic is implicitly xkcd:topic (ie main:YouTube discusses the xkcd comic, while ref:YouTube is the place where the pop-culture reference of YouTube is discussed.) Either that, or some other name decoration, such as YouTube Explained, or ... -- IronyChef (talk) 13:59, 7 August 2012 (UTC)
Agreed. Number and the name together. --Jeff (talk) 16:08, 7 August 2012 (UTC)
Looks like we have consensus. I'll move the pages (I've been meaning to learn how to use mwclient anyway :D) --Waldir (talk) 18:01, 7 August 2012 (UTC)
 Done, all current pages have been moved. However, I am not sure whether we should keep a space after the colon. What do you guys think? Should it be "112: Baring My Heart" or "112:Baring My Heart"? --Waldir (talk) 18:20, 7 August 2012 (UTC)
Also, I just realized MediaWiki doesn't allow colons in image Filenames. One solution could be using something like File:786. Exoplanets.png or File:786-Exoplanets.png, but then perhaps we'd have to change the pages name too, for consistency? I'll try to investigate what is the reasoning behind this restriction. --Waldir (talk) 18:50, 7 August 2012 (UTC)
Ok, it seems like it's a matter of setting $wgIllegalFileChars = ''; in LocalSettings.php (because it is set as $wgIllegalFileChars = ':'; in DefaultSettings.php). Jeff, could you do that please? --Waldir (talk) 19:13, 7 August 2012 (UTC)
I guess this is my bad for not ciming in on this discussion earlier, but I frankly think that the #: Name is a worse way of doing it just for the reasons of system resources. #:Name is fine from a user standpoint with the caveat that # and Name both redirect to #:Name. The problem is that this requires 2 redirects minimum for every comic, and the redirect itself takes a bit more time for each article to load, and (as I understand from wikipedia and its dislike of double redirects), every redirect adds to the system load. So if every article lookup by users (who will undoubtedly type either the number or the name, but rarely both) is a redirect, the system load is going to go up.
As an aside, assuming Jeff is able to install the Cite Extension to add citation referencing (and even if he doesn't), I was expecting to try to create some sort of template in the concept of {{cite comic}} where you could basically pass a single variable (e.g. the comic number) and it would create a proper citation for that comic. Similarly, this naming format will perhaps require a template something like {{comicno}} with a comic number field just to create a quick link that is visibly appealing and links properly to the comic with that number. (ie: {comicno|18} would produce a link like "Snapple" or something). I'm wondering though if anyone has any coding ideas for how we might accomplish this other than the hardcode all the titles into a template. TheHYPO (talk) 19:26, 7 August 2012 (UTC)
PS: I did some mild digging on another wiki, Star Trek's Memory Alpha wiki, and although all of its episode articles are now titled "episode title (episode)" to avoid disambiguation, which allows you to an episode template by calling the title (which template appends "(episode)" to every entry), they DO have a title-display template: Template:Titles - with a template subpage for every single episode setting out how the mouseover text should be displayed. It would be possible to do such a template for xkcd just so that comic numbers can be crossreferenced to titles... TheHYPO (talk) 20:30, 7 August 2012 (UTC)
(Hoping this is the right number of colons for proper indentation... ;-) Redirects are one thing, and while probably resulting in possibly two page serves (isn't it really just two hits to the db?) they're natively supported by mediawiki. Even so, if performance is proven to be a real (not just conjectured) problem, can we do something clever, perhaps, with transclusion? Either the number transcludes the title, or vice versa? Might be a case of pre-optimization, though; in the back of my mind, it seems that the rendering engine puts as much effort into transcluding to expand templates as it would to expand a redirect in situ: either case is just a query to the DB to expand the contents of said item. (Enough rambling; anybody have any concrete metrics on this?) -- IronyChef (talk) 06:23, 9 August 2012 (UTC)
Hi folks. Just thought I'd state that redirects are completely safe. They don't add any noticeable loading time for the users and the extra resources used by the server are so minor that it's akin to the resources used to type a character in notepad. Pages are also aggressively cached (by default, anyway). If you're interested, the way redirects work in Mediawiki isn't like most other sites handle redirects. It's not loading a page that makes you load another page. Rather, all content is stored in an SQL database. The content is stored under a certain name (eg, "#: Hello World!"). A redirect simply tells Mediawiki to look for the content under a different name. Slightly more work for the server (don't worry, they can handle it), but the page is delivered to the user in roughly the same period of time (if we want to be technical, the page will be slightly larger, due to the "Redirected from whatever" line added to the page (which is mostly there for the purpose of making it easier to fix incorrect redirects). I don't have metrics, but can assure you that it's almost no difference in the end result. Omega TalkContribs 09:11, 9 August 2012 (UTC)

I've been thinking about this some more, and I believe we should choose a different pattern for the page names.

  • First, use another separator between comic number and name, since colon is forbidden in files. A simple alternative would be "Comic title (number)", as in Michael Phelps (1092). This would additionally allow us to use the pipe trick when linking to a comic, since content in parenthesis is automatically stripped out: [[Michael Phelps (1092)|]] results in Michael Phelps. Another effect of this is that by dropping the colon naming scheme we would remove ambiguity with the namespace system, which also uses colons to separate namespaces from pagenames.
  • Second, we should probably follow IronyChef's suggestion above and move them to a specific namespace, such as Comic:Michael Phelps (1092). Other namespaces could be added for more topics, such as Character:Cueball, xkcd:Randall (or Meta:Randall), Topic:Velociraptors, etc. Not only we would be able to generate lists of pages without resorting to categories (which have to be added manually), but we would get lot's of "Random X" for free (random comic, random character, random topic, etc.)

What do you guys think? --Waldir (talk) 14:29, 9 August 2012 (UTC)

P.s. - Proper category sorting of the comics would be dealt with by the {{comic}} template, which would also pad the numbers with zeroes to ensure 100 comes after 2, etc.
+1 on the parens... (but does that mean my recent double-redirect-fixups have been for naught? (grin)) ... I couldn't put my finger on it and didn't articulate it earlier, but the fact that colon needed special attention by the software left me a bit uneasy (there must be a reason for them doing that, like namespaces perhaps) so using parentheses-es-es (as long as we close them properly) seems more the mediawiki way. -- IronyChef (talk) 15:03, 9 August 2012 (UTC) (I know you folks don't like my propensity to (over?)categorize, but [[Category:Parentheses]] is just too irrisistable... ;-)
I think, that all of this seem unnecessary complication to me. I don't see any problem with the current system. I think something like 1092: Michael Phelps flows well, is quite readable and easy to insert "as is" in the text (see the links to other comics in 1048: Emotion for instance). As I understand, we would want the image files to be titled exactly the same way as their corresponding article; why, where is the need for that? (to me the simplest way, and most relevant maybe, would be to name them exactly as they are on xkcd.com; maybe with a prefix, like "xkcd - ", so that it cannot mess with other existing images such as from Commons).
I don't see the point of creating namespaces such as "Character", "Topic", etc.; what is the problem with Beret Guy, Randall Munroe, Velociraptors, and such? with namespaces one will have to put each topic in one box (and one only), where will you put things like Stick figure or My Hobby or any other thing that will pop up without clearly belonging to one of these boxes? just give up! :-)
About the "Random X", I like the idea that on xkcd.com, you can get a random comic (because that's all what is there), but in here you can get a random whatever: you may get a comic explanation, a character, a topic or anything, because in here there is all that.
I don't think the colon in the comic page names will pose any problem, it cannot mess with anything as long as it is preceded by a number only.
In the end, I think that adding the number in the comic page names was a good choice, because there would have been real issues otherwise, but for now I would say : "don't fix what is not broken", KISS, and "just give up". :-)
Cos (talk) 16:14, 9 August 2012 (UTC)
I have to agree with this. The existing page names are fine in my book, and I don't see any benefits of renaming them all (again). Concerning the random, though, I mentioned an extension in proposals that would allow us to choose a "random page in a category". I don't really care one way or another about character topics. Seem like a lot of maintenance when we don't even have a quarter of the comics explained yet, but whatever. Concerning the image names, I think that simply using the same name as it appears on xkcd is fine. Images are a bit of a "backend", that people don't usually search for (rather, they'd search for the comic and find the image on that page). As well, since all images are hosted on xkcd, they won't be any file name conflicts amongst the comics. Omega TalkContribs 18:04, 9 August 2012 (UTC)