Wikipedia:Village pump (technical)/Archive 14

New "nbsp" below the edit box

The code &nbsp; has recently been added to the resources below the edit box, in the section "Wiki markup", just between #REDIRECT [[]] and <s></s>. That's great!

I'm with a group of editors who are currently interested in the hard space: its markup and its uses in editing. Some questions:

  • Exactly when was that change made?
  • Who made it?
  • Where are such changes logged?
  • What is the very best place to propose such changes? (Seems to be WP:VPR; am I right?)

Any authoritative information on these will be of great value.

– Noetica♬♩Talk 01:31, 8 January 2008 (UTC)

It's at MediaWiki:Edittools. You can make suggestions on the talk page. &nbsp; was added yesterday.[1] PrimeHunter (talk) 01:41, 8 January 2008 (UTC)
Many thanks, PrimeHunter!
– Noetica♬♩Talk 01:44, 8 January 2008 (UTC)
All jolly good - but what does it do? DuncanHill (talk) 01:59, 8 January 2008 (UTC)
If clicked while the cursor is in the edit box, it inserts the html character code &nbsp;, the code for a Non-breaking space, at the cursor position. -- Boracay Bill (talk) 02:35, 8 January 2008 (UTC)
Thanks. DuncanHill (talk) 02:40, 8 January 2008 (UTC)

Is there a centralized bibliographic database for wikipedia? Is there a way to make citations just by giving an universal ID instead of copying a full citation template?

Is there a database which collects all {{citation}} used in wikipedia so that we can reuse the references that are already registered from there?

More generally, is there a mechanism à la BibTeX which would allow to give just a reference identifier in an article and that the corresponding reference in the global database is automatically used.

The Reference Wikification tool is a very good tool to automatically convert Google scholar entries into Wikipedia citation entries but does there exist some way to just give an identifier (e.g. the DOI, ISBN, or the canonical BibTeX name used, for computer science, on DBLP) and that the reference is inserted automatically, without having to copy-paste any kind of citation template in the article, in the same way it is done for LaTeX articles using BibTeX? Hugo Herbelin (talk) 14:02, 29 December 2007 (UTC)

No, no, and no. Sorry. MilesAgain (talk) 14:04, 29 December 2007 (UTC)
There is a database that some editors in the mathematics project use. It is maintained by User:Jakob.scholbach at http://zeteo.info . It is only intended for math articles, though; I don't know the resource limits on that server. — Carl (CBM · talk) 14:55, 29 December 2007 (UTC)

This seems like an interesting project to pursue. But where to get funding? Or: perhaps there are already mediawiki plugins that allow this? (Hmm, perhaps I could modify Omegawiki to cover this kind of thing... brrr....) --Kim Bruning (talk) 14:59, 29 December 2007 (UTC)

Such a mechanism is mentioned in a collection of related items in WP:FOOT. Wikicat/Wikicite is slowly being worked on. -- SEWilco (talk) 15:03, 29 December 2007 (UTC)

On my private wiki, I used a "Reference:" namespace to collect citations and then just transclude from there, i.e. {{Reference:Taylor et al. 2005}}. We could plausibly do something similar, though we'd probably have to come up with a better naming scheme to avoid conflicts. Dragons flight (talk) 15:23, 29 December 2007 (UTC)

URL stands for Uniform Resource Locator. If there was a URL: namespace, then someone could enter URL:specificurl into the search box, and either get a page with a complete citation, or a page with a preformated but blank (except URL) citation template that could be filled out, then copied and pasted. A bot could monitor new edits, and when it saw a citation added to an article, could create a new page in the URL namespace if such a page didn't already exist. Or, when a naked URL was added to an article (no citation), and the page existed in URL namespace, the bot could replace the naked URL with a full citation.
An ISBN namespace would be even better - not only could the citation info be populated automatically (say, from the Library of Congress database), but editors could post that they owned the book and could assist other editors interested in getting info from that book. -- John Broughton (♫♫) 16:17, 29 December 2007 (UTC)
ISBNs are a type of URN. (urn:isbn:0-395-36341-1) That allows you to cite books, ISSN allows you to cite entire journals, and SICI allows you to cite articles within a journal. (urn:sici:1046-8188(199501)13:1%3C69:FTTHBI%3E2.0.TX;2-4) I suggested this a while ago, but it was considered too complex and confusing. Really we need a dedicated user interface for adding references that doesn't depend on cryptic code (citation templates or otherwise) within the article at all. — Omegatron 05:51, 30 December 2007 (UTC)
I'd like to comment further on Jakob.scholbach's tool called Zeteo that was already mentioned. For an example, click on this link and it opens up an editing screen for a journal reference to a mathematical paper by Vladimir Chernousov and Jean-Pierre Serre. In the box called 'citation' at the bottom of the screen there is a {{Citation}} template for that reference that is ready to cut and paste into an article. Scan through User_talk:Jakob.scholbach for some reactions that other editors have given to his scheme. Nothing limits the use of his tool to the mathematical domain.
The Reference wikification tool mentioned above by User:Hugo Herbelin also produces WP citation templates but as a search result. So you do a Google Scholar search, find the paper you want, and click on a special link to generate the citation template. (This is different from saving all the citations in a central database a la Scholbach).
User:Diberri has a tool at http://diberri.dyndns.org/wikipedia/templates that will convert an ISBN number into a properly formatted citation template that includes the book author, title and publisher, if all the lookups succeed. It also converts a Pubmed ID into a citation template for a journal reference. For me, this tool provides a good reason to include Pubmed IDs as part of every journal reference when working on any biological article that is related to the PubMed domain. To see how this works, click on the PMID in the following reference:
Carrington J, Ambros V (2003). "Role of microRNAs in plant and animal development". Science. 301 (5631): 336–8. PMID 12869753.
Diberri's system and the Pubmed system are 'bidirectional', in the sense that they use an ID number that is actually included in the Wikipedia citation. Jakob's tool is bidirectional for books (since it can look up ISBNs) but not for journal articles, since it doesn't give you a clickable reference ID number. If you are looking at a Wikipedia reference and want to know if it's recorded in his system, you need to do a lookup by author or title.
When I first saw Jakob's proposal it reminded me of http://www.usin.org. An article link in that system could be like [2] for a book, and there is also supposed to be a lookup for journal articles. Of course there are also DOI links and JSTOR links that some online systems already understand. There is a list of existing tools at Wikipedia:Citing sources#Tools. The French Wikipedia already has a central system for citations of math textbooks that are used in more than one article (example at [3]). Other interesting tools are CiteULike, Zotero and LibraryThing. The latter database will, if given an ISBN, return the list of articles in Wikipedia that mention that ISBN. EdJohnston (talk) 18:40, 29 December 2007 (UTC)
Thanks for all these very interesting informations. Would it be relevant then to have a more detailed presentation of the different tools and of their usefulness for dealing with citations? On the Wikipedia:Citing sources page there is only a list of them but no explanation on their recommended use. —Preceding unsigned comment added by Hugo Herbelin (talkcontribs) 22:26, 29 December 2007 (UTC)
If I can do anything to help the wiki community with this task, I'm committed to it. As was mentioned above, zeteo is not at all limited to mathematics. I'm currently running the http://zeteo.info website at my personal cost, but at the moment there is pretty little traffic. So, there are no a priori obstacles in extending the usage to non-mathematical stuff. It was a somewhat tedious work to feed the existing templates into the database (see also this thread), but if it is carried by several shoulders, should be feasible. The some 20.000 math articles (I think) contained some 5.000 reference templates (journals and books only).
Any other reference obviously fits into the database, too. This may be a stupid question: is there a commonly used unique ID for papers in journals? (comparable to the ISBN vis à vis books). I could think of glueing together ISSN, issue and volume, but I'm not very much into these bibliographical things. But actually, perhaps it would be even better to have something like
{{reference database|query=Grothendieck points}} or {{reference database|isbn=978-0-691-08238-7}}
which could point to [4], [5], respectively (or to a similar different database). These unique IDs like ISBN and the like are nice, but for everyday editorial work they are (I think) less useful, because one does not know an ISBN by heart, but one usually does know the author and the title. Jakob.scholbach (talk) 01:03, 30 December 2007 (UTC)
is there a commonly used unique ID for papers in journals? (comparable to the ISBN vis à vis books).
Yes: SICI. It is used in JSTOR URLs, for instance. — Omegatron 06:04, 30 December 2007 (UTC)

OttoBib added Wikipedia citation format a while ago[6], but zeteo.info looks much more advanced and has a better "understanding" of Wikipedia templates and citation needs.

The "Science Online" wiki has an enhanced citation system which allows citations to be defined globally, allowing bidirectional navigation of bibliographic data. See http://www.scionline.org/index.php/Special:ReferenceBrowser and http://www.scionline.org/index.php/Citing_sources . Note that they also have a "Reference" namespace, which is a standard page with associated talk page, but has a "reference" block at the top with its own edit interface. It does not support many of our commonly used params like ISBN, ISSN, DOI, OCLC, etc. but they would not be hard to add.

I think we should follow the French approach and create a "Reference" namespace. Here is another example: fr:Référence:Physique théorique (Landau et Lifchitz). The French namespace is almost two years old (creation proposal: fr:Wikipédia:Prise de décision/Espace référence), has a wiki project to manage it (fr:Projet:Sources), and is receiving new pages at a steady pace with 878 pages currently in the namespace. In addition to time saving benefits for sourcing, the talk page will be used for transcriptions, translations, and requesting verification (see Wikipedia:WikiProject Fact and Reference Check), and PD sources which are often used will be easier to find in order to transcribe onto Wikisource. John Vandenberg (talk) 05:37, 30 December 2007 (UTC)

There should be. References should be added through an interface that's not part of the article source code at all. (And so should categories, interwikis, etc.) — Omegatron 05:45, 30 December 2007 (UTC)

A word on the french system: I believe it is unusable when it comes to bigger amount of data (with my excuses to the creators of this system). In order to compare the situations between fr.wp and en.wp, one should notice that the French WP does not have parallels of {{cite journal}} or {{citation}}, and the overall referencing density also seems to be lower. Currently, citing journal articles is simply done there by hacking it into the articles, which is probably the least efficient/stringent method.
At least half of references are journals (I can speak only for the math articles: some 2.600 reference items in zeteo are journal papers, about 1.600 have ISBNs, so the status of the remaining 1.100 is hard to decide with a single SQL query). I guess, books are in general more often in several WP articles. As I tried to show with the setup of zeteo, it is not only necessary to have structured information about books and journal papers, but also about journals (their ISSN, for example), publishers (location), authors.
We want/need to handle an eventually huge amount of bibliographic information. Therefore it is, I'm convinced, critical to have a structured pool of data. Only a database in a more proper sense of the word can accomplish this. Issues like discussing an item's correctness, grouping together translations and several editions of a work are definitely necessary (and not yet proposed by z.), I agree.
  • Are there any Wiki-style databases?
  • Is it possible to set up such a database inside WP (as opposed to an external website)?
  • Or differently put, the whole discussion raises the question, whether WP as it currently is, lacks the opportunity of providing structured information and what we can do about this. Jakob.scholbach (talk) 12:47, 30 December 2007 (UTC)

A couple of comments; first the current referencing model makes some wikisource almost unreadable, secondly many books have multiple ISBNs. Rich Farmbrough, 14:09 30 December 2007 (GMT).


I dont see how the increased size of the English Wikipedia makes the French facility any less viable. The problem domain remains the same; enWP is just a bigger implementation.

While this approach would be comprised of standard pages, the French pages in this namespace are very structured: the pages consist almost entirely of entries constructed with templates. I suggest we improve on the frWP solution by making each Reference page also function as a citation template, and by adding an alias from REF to Reference. For example, referencing s:Real Soldiers of Fortune would be as simple as {{REF:Real Soldiers of Fortune (Davis, 1906)|chapter=3}} and the template could automatically display the correct chapter name. The template could also flag an error when called without a chapter within the appropriate range.

This approach works equally well for journals. Reference:AN could redirect to Reference:Astronomische Nachrichten which would have a template of :

<includeonly><ref name="AN-{{{1}}}:{{{2}}}-{{{title}}}">{{cite journal | journal = [[Astronomische Nachrichten]] | volume = {{{1}}} | issue = {{{2}}} | title = {{{title}}} | .... }} <sup>([[Reference:Astronomische Nachrichten|more info]])</sup></ref><includeonly>

which could be called on any page using:

{{REF:AN|1|1|title=Vorwort}}

The same reference could be repeated in that format on the page many times and the Cite engine will automatically match them up because the <ref> "name" parameter is the same. Each "Reference" page could tailor its own template to suit the intricacies of that specific work and the inbuilt assumptions that can be made in the case of that work. The following could all work (note that we already have specialised cite templates for many oft used reference works; all I am suggesting is that they could become a standard part of the new Reference: page design):

{{REF:EB1911|Churchill, Lord Randolph Henry Spencer}}
{{REF:WFB|Somalia|year=2004}}
{{REF:NYT|1905|12|30|title=BOSTON NOTES. Some New Fiction - A Musical Dictionary by Nathan H. Dole--Some New Editions and Other Gossip About Books.|author=Stephenson Browne}}
etc, etc.

This would not replace the existing {{cite}} templates; these "Reference" templates would call those {{cite}} templates in a consistent manner. A namespace would allow people who often use the same reference material to streamline their editing, and would allow the bibliographic data to be mined more readily.

m:Wikicat is the wiki-style structured database Extension that provides the desired functionality inside MediaWiki, and it is moving ahead, with recent checkins, however I think a "Reference:" namespace is a simple interim approach that will bring additional minds to bear on how this beast should end up working, and will increase the developer base working on Wikicat. Even with a structured datastore added to Wikipedia, the Reference namespace will still be a useful layer on top, allowing unstructured content and discussion to occur.

mw:Category:Referencing extensions has other referencing extensions that may be of interest. John Vandenberg (talk) 15:30, 30 December 2007 (UTC)

m:Wikicat is particularly interesting: an early step in the implementation process is The Cite extension is enhanced to support automatically fetching bibliographic header information when only a key (i.e. ISBN number) is provided to the Cite reference tag. The imported bibliographic data replaces the key within the article wiki-text ... So if the SICI were able to be copied and pasted from articles (easily), that would go a long way to automating citation information.
And, at the risk of offering yet another option, it seems to me that Wikipedia - and the desirability of links to pages with advertising - have gotten so large that it shouldn't be that hard to persuade major newspapers to add XML or other markup to their stories so that a browser (add-in) can immediately generate a proper citation. With that in place, it would only take a couple of clicks to have a footnote ready to go. -- John Broughton (♫♫) 16:17, 30 December 2007 (UTC)
And, poking around, I came across Biblio.php (alternative site), which also pulls in citation data from external databases; it's actually operational on a MediaWiki website. And a two-year-old bugzilla feature request - "A reference system that support BibTeX databases".
Given all that, this seems to be the sort of thing where a bit of an organized effort is needed. There are three sort-of-related Wikiprojects: Wikipedia:WikiProject Citation cleanup, Wikipedia:WikiProject External links, and Wikipedia:WikiProject Fact and Reference Check, but I don't think any of them are focused on this. -- John Broughton (♫♫) 16:37, 30 December 2007 (UTC)
mw:Extension:Biblio looks quite good. My only reservation is that it uses a duplicated set of tags: <cite> and <biblio/> are functionally very similar to <ref> and <references/>, and there isnt any explanation on how they can co-exist. i.e. would "<references/><biblio/>" result in two numbered lists rather than one merged list. John Vandenberg (talk) 04:28, 31 December 2007 (UTC)
I've spammed those WikiProjects and more (Resource Exchange, Librarians‎, and Wikicite) to gather more heads together on this. John Vandenberg (talk) 04:28, 31 December 2007 (UTC)
(edit conflict) Replying to "I dont see how the increased size of the English Wikipedia makes the French facility any less viable": I just think the more items you get, the more difficult it is to handle them. Look at the oeuvre of Sophocles, for example. The system / the editors created templates called {{SopAja}}, {{SopŒC}} etc. I guess these names have been chosen by an editor. Sop stands for Sophocles etc. Just imagine you have 20 other authors Sop***. Then you would need to extend the current 3-letter scheme to more letters etc. Or imagine a book written by 5 authors. Or take a book written by Bloch. (André, Bernard, or one of the other 20 Blochs?) If these templates are generated manually, the result will be a mess, because every editor will choose his favourite scheme of encoding author's name, title, possibly journal etc. into a unique identifier. If editors are restricted to use a certain encoding scheme like
{{REF|author1name_author1firstname_[author2name_author2firstname...]_title_[journal]_[isbn]_[year]}}
we end up at something close to {{citation}}.
I like the above template idea. The more flexible the underlying storage system is, the more flexible the admissible {{REF|...}} templates proposed above can be (and hence the less repelling it is for editors to add a reference). A template of the kind {{REF|Bloch, Freiheit und Ordnung}} which would point to an engine looking up works by an author called Bloch and title F.u.O., thus retrieving this work of Ernst Bloch (or if the information is incomplete, would retrieve several books etc., or also several editions) is better than {{Bloch_Ernst_FuO}} or similar things. It is rare (never occurs?) that there are two different works (not two editions etc.) having the same last name of some author and the same title. Therefore a template like {{REF|author, title}} should be fine. If some duplicates of this kind do arise, they could still be distinguished à la {{REF|author, title, year}} etc.Jakob.scholbach (talk) 16:54, 30 December 2007 (UTC)
It is important not to loose sight of the problem: making it easier to regularly cite the commonly used works.
Works that are rarely referenced are not the target audience, and editors which dont regularly use the same work are unlikely to create these Reference pages. The management of this namespace would have similar restrictions to the Template namespace. e.g. a reference only used on one page is subst'ed and then deleted.
A Reference namespace allows a higher level citation specification language to be developed by the community, to reside on top of the existing {{cite}}. The management of this language would be primarily self-governing, with topical lists of references being created by WikiProjects. e.g. fr:Projet:Mythologie grecque/Modèles de sources. I agree that automatically creating Reference: pages would not be advisable.
  • {{REF|Bloch, Freiheit und Ordnung}} requires a MediaWiki Extension that does not exist and, if the slow progress of m:Wikicat is any indication, will not be available for at least another year, and will depends on programmers to develop solutions to deal with the many complications of addressing works. It also depends on importing from external databases, many of which are non-free (isbndb, oclc, etc).
  • {{REF:Bloch, Freiheit und Ordnung}} can be set up right now, uses PAGENAME as the unique identifier, and has the additional benefit of being "written" in a language that every Wikipedian understands and can modify.
John Vandenberg (talk) 02:19, 31 December 2007 (UTC)
Well, if you think this is a good way, go ahead. Perhaps have a talk with the French users. I've talked, [7] only to one (who was not exactly in favor of it). As an aside: the vast majority of reference items does not occur more than once (in numbers, always referring to the math articles, which were pulled out by User:CBM: 6.400 items in total, 5.250 of which occurred once (counted by comparing the title only), 470 twice, 102 thrice and about 100 occurred more than thrice). So it is only the tip of the iceberg which you would adress with the above idea. As you indicate you would delete a subpage for a reference which is just used once, which seems to be pretty much repelling. But if you want to start with, say, the 100 most frequent ones, which could perhaps be extrapolated to some 10.000 for en.wp as a whole, this can be helpful. Probably it is even relatively easy to pull the stuff out, and in some semi-automatic way create all these reference subpages. But it is, as we say in German, only a drop on the hot stone.
I can only offer to participate in writing a database, which could fill the gap before (if ever) there is the wished extension of mediawiki. Another thing I can offer is to publish the source code and/or the contents of zeteo in some way so that more people can have a look/try to improve it along the lines discussed above. I wrote this database (just learning PHP/SQL along the way) in short time. Importing the math articles templates was a matter of a couple of nights (stupid clicking all the time, though :-)). So I think with relatively little effort the result can be much better than with the subpage approach. Jakob.scholbach (talk) 15:24, 31 December 2007 (UTC)

Just the fact of being able to point to a citation would be so dang useful in so many levels. We have just divided the list of best-selling video games article into two different ones, and considering the original had 214 references reviewed and kept up to date for over 100 games (each with its own article), it would make life (and ours in particular) much, much easier. And if you add a "What links here" function to every citation, it would make life even much simpler! -- ReyBrujo (talk) 04:57, 1 January 2008 (UTC)

Category:Specific source templates are pre-rolled {{cite}} templates. By creating one of those, whatlinkshere will work. see Special:Whatlinkshere/Template:Greenwood&Earnshaw. John Vandenberg (talk) 11:09, 3 January 2008 (UTC)
  • It is interesting to reflect on how reference templates might be parametrized, and how many different places a given template might be used.
  • See Template:Abramowitz Stegun ref which is used in math articles, and can be invoked with specific chapter and page numbers, since different sections are cited in different articles. A redirect allows it to be accessed as Template:AS ref. Depending on the parameter values, it generates separate URLs for the various chapters, due to the complexity of the web site that hosts the convenience copy of this work. A number of the templates in Category:Specific source templates take parameters, though the syntax varies.
  • John's example of Template:Greenwood&Earnshaw is invoked from 89 articles but without parameters, so it always expands to the same thing. Template:AS ref is invoked from 14 articles but expands to different things, sometimes to two different things in one article.
  • Nothing stands in the way of any WikiProject that wants to go ahead and create either fixed or parametrized templates for commonly-used works. (No new software has to be written, just some templates).
  • This won't help much for the example just given by ReyBrujo where a list article about video games had 214 references, some of them being shared with the articles on the individual games. I assume that each reference would be used at most twice (once in the article for the game, and once in the List article summarizing all the games). EdJohnston (talk) 02:38, 6 January 2008 (UTC)
If you want I can make zeteo (optionally) generating a source code for such a template which would then only have to be pasted to a template subpage. For example, "The value of science" is quoted some 20 times in math articles. The user could save that a template with a certain name {{xyz}} was generated for a given reverence and optionally display the template {{xyz}} instead of {{citation}}. Jakob.scholbach (talk) 20:19, 6 January 2008 (UTC)

As far as WikiProjects using reference templates goes, have a look at {{ME-ref}}, {{ME-cite}}, and {{ME-source}}. User:CBDunkerson did those for Wikipedia:WikiProject Middle-earth. Carcharoth (talk) 17:45, 8 January 2008 (UTC)

Transparent background in <math>

The pure-white background that appears in when <math> formulas produce a PNG image is a problem, for example in wikitables where the background is grey enough that the white stands out.

like this:  

Is there any way to have <math> produce images with transparent backgrounds instead? MilesAgain (talk) 09:24, 1 January 2008 (UTC)

I don't think so. You'll have to go to Bugzilla. (In the bug description, it might be handy to link to m:Help talk:Displaying a formula#Maynard Handley's suggestions). MER-C 11:31, 1 January 2008 (UTC)

Could someone with bugzilla experience please do this? MilesAgain (talk) 13:31, 8 January 2008 (UTC)

It's possible, but we don't do it for two big reasons:
  1. Unlike body text, the rendered math foreground color will not change along with system or browser preferences (such as someone with poor vision might choose to improve high-contrast legibility). If the selected colors use a dark background, the text will be illegible.
  2. In some browsers, transparent images print with a black background, making the math text illegible.
No need to open a new bug; this is bugzilla:8 and closed as WONTFIX for several years due to the reasons above. --Brion VIBBER (talk) 22:13, 8 January 2008 (UTC)

Interwiki into extra namespaces

Spanish Wikipedia decided to create a new namespace for Lists. So bot is now removing interwiki links to those lists. To me this "not a main namespace" argument looks like a formality in this case, any other opinions? See discussion at ia:Discussion_Usator:Synthebot#00029 and sorry if this has been deiscussed before ∴ AlexSm 16:18, 7 January 2008 (UTC)

That's a poor judgement call. Anexo namespace has also encyclopedic content. Interwikis serve a function to help locating same information across languages. If the corresponding information from here is located in a new "non-discussion" namespace at other wikis, the interwiki must be preserved in order to help locating the information. -- drini [meta:] [commons:] 16:36, 8 January 2008 (UTC)

Leftover image thumbnails on the servers

I saw an image in the log that was deleted last week (Image:Architecture of CRM.JPG). Yet the popups thing displayed a thumbnail of it when I hovered over the link. I copied the image url and noticed that some different thumbnail sizes of the image are actually still there. E.g., [8]. Is that supposed to happen? I thought deleted images would have to go immediately, in case of things like copyright infringement or stuff illegal where Wikipedia's servers are hosted. • Anakin (contribscomplaints) 00:57, 8 January 2008 (UTC)

Not really answering your technical question, but the critical thing for improper images is to get them out of articles, so that (a) readers don't see them) and (b) all of the Wikipedia mirrors don't copy them to those other websites. That takes care of the copyright concerns. -- John Broughton (♫♫) 14:47, 8 January 2008 (UTC)

help with Template:Infobox Mountain

Hello. I was wondering if there is anyone out there who knows how to add in an image resizing option into Template:Infobox Mountain. The current template forces image size, which is fine for horizontally aligned images, but inflates vertically aligned images too much. Its parent, Template:Geobox, allows for a resizing option. I've brought this to the talk page for the Wikipedia:WikiProject Mountains; concensus exits for the template change, but no one seems to know how to do it.--Pgagnon999 (talk) 03:38, 8 January 2008 (UTC)

Your problem was already solved a couple of days ago by Circeus. You can now use the option {{{Photo size}}} to specify a size for non-landscaped photo's. --TheDJ (talkcontribs) 13:51, 8 January 2008 (UTC)

Query

Is there a quick and easy way of working out the head count of admins in the support and oppose sections of the poll at Wikipedia:Non-administrator rollback/Poll? Just interested in the raw stats. Hiding T 23:16, 8 January 2008 (UTC)

You want to know the totals for just the admins, right? Don't think there is a way to do that, other than manually. - Rjd0060 (talk) 04:40, 9 January 2008 (UTC)

Minor diff visualization problem

 
diff error

Anyone besides me notices something wrong with this diff? Maybe the Unicode and ASCII mixture makes the red word appear overlapped? -- ReyBrujo (talk) 18:31, 1 January 2008 (UTC)

I see nothing wrong in Internet Explorer 7.0. It sounds like a browser problem. PrimeHunter (talk) 22:46, 1 January 2008 (UTC)
Shows fine in IE6 as well. EdokterTalk 00:02, 2 January 2008 (UTC)
No issue in Firefox 2.0 either. Do you have complex script support installed? —Remember the dot (talk) 00:07, 2 January 2008 (UTC)
I see no problem with either IE 6.0 or Firefox 2.0.0.11 -- Boracay Bill (talk) 00:52, 2 January 2008 (UTC)
Happening in Linux Ubuntu 2.0.0.11. I am putting an image here. I have never had this problem before. -- ReyBrujo (talk) 01:57, 2 January 2008 (UTC)
Doesn't happen under Windows, so I guess it is just my configuration. Unless it is for Linux too (nobody with Linux to try that out?) -- ReyBrujo (talk) 16:10, 2 January 2008 (UTC)
Happens to me on Ubuntu Gutsy Gibbon running Firefox 2.0.0.11, just like the screenshot. 1 != 2 04:08, 3 January 2008 (UTC)
Weird, I see nothing suspicious on the HTML code, that's probably a bug either in the Linux version of Fx, or your Desktop Manager. -- lucasbfr ho ho ho 12:46, 3 January 2008 (UTC)
I'm on Ubuntu Gutsy Gibbon / Firefox/2.0.0.11 at the moment, and I see the same view as the screenshot above. --ais523 18:33, 3 January 2008 (UTC)
Apparently it is a Firefox under Linux, unless someone using Gentoo or another distro can confirm it is not happening to them. -- ReyBrujo (talk) 01:36, 4 January 2008 (UTC)
Doesn't happen on Macintosh either with Safari 3.0.4 or Firefox 2.0.0.11. EdJohnston (talk) 01:23, 6 January 2008 (UTC)
I am running Ubuntu Feisty with Firefox 2.0.0.11 installed, and the diff linked to in the OP here appears just fine. --MikeVitale 14:33, 9 January 2008 (UTC)

Impossibile size

I did some investigations to see if porting wikipedia to mobile device is possible, and I found incredible, impossibile results!!!

Size of italian wikipedia dump ( http://download.wikimedia.org/itwiki/20071215/ ): 418 MB zipped

Size of english wikipedia dump ( http://download.wikimedia.org/enwiki/20071218/ ): 760 MB zipped

Being it just text (no images, no history, no comments, just data), we can suppose a 10:1 compression ratio.

So we should have around 4 GB for italian wikipedia.

I read Encyclopedia Britannica sizes around 1 GB, and this is already weird... But that's not all:

supposing 500 pages volumes, 2 columns per page, 300 rows per column, 40 characters per row, we have 12,000,000 characters per voume. 4 GB = 4,000,000,000 characters, i.e. 333 volumes, i.e. 30 feet: italian printed wikipedia requires a 30 feet shelf?!? english wikipedia requires almost 60 feet shelf?

That's obviously not true. There's something terribily wrong in wikipedia dumps.

What's going on?!?


additional note: a 20-volumes encyclopedia would require around 200 MB (without images).


--Cassioli (talk) 09:45, 7 January 2008 (UTC)

Actually, you are looking at a broken dump. Current EN articles compressed is ~3 GB. So far as being enormous, well, we are. Encyclopedia Brittanica has a mere 40M words. EN is now somewhere around 800M, so yes, enwiki would fill roughly 20 of your 20 volume encyclopedias. We have more articles and substantially more words per article. Where else would one devote 2300 words to spoo? Dragons flight (talk) 10:14, 7 January 2008 (UTC)
See Wikipedia:Size of Wikipedia and Wikipedia:Size in volumes for some more crazy figures —Preceding unsigned comment added by Graham87 (talkcontribs) 13:09, 7 January 2008 (UTC)
Bear in mind most online encylopaedias are larger then their printed companions (if any exist). That might be why Encylopaedia Britannica online is 1 gb (if your figure is correct) whereas the printed version is only about 200 MB Nil Einne (talk) 08:37, 9 January 2008 (UTC)
In attempting to port Wikipedia to a portable device, remember that most of it can be dumped. See commons:Image:Size_of_English_Wikipedia_broken_down.png. That image is satirical and exaggerated, but seriously, there are some articles no practical everyday reference guide should ever need. Like articles about high schools that nobody who didn't go to them would care about, point-of-view copyvio junk, and hundreds of pages of episode guides of obscure anime. The Wikipedia 0.5 release included only a very carefully chosen 1964 articles out of the 2 million in the encyclopaedia, and as a result, the usefulness density (ratio of junk to useful stuff) is *far* higher.... Using only select articles from Wikipedia I'm sure you could create a very valuable reference that would fit in a couple of hundred megabytes. • Anakin (contribscomplaints) 16:24, 9 January 2008 (UTC)

Automatic clean-up notifications

I'm not particularly sure I'm even asking in the right section here, but I figure I'll get pointed in the right direction anyhow! I was wondering if it's possible to generate an automatic list of articles with different cleanup issues, say for a WikiProject? At present, updating a to-do list consists of browsing through articles and then adding articles with issues as you come across them. It'd be great if there was a way to automatically generate to-do lists from a list of related articles, such as those in a certain WikiProject. Hope that makes sense! Thanks in advance, Seaserpent85 23:42, 8 January 2008 (UTC)

I think this has been discussed before, and I don't think there is a way to get that type of list of related pages. I could be wrong though. That would really be a useful tool though. - Rjd0060 (talk) 04:39, 9 January 2008 (UTC)
The only solution I can think of is to keep an eye on the categories that an article is placed in when a certain cleanup template is added. This doesn't generate separate do-to lists, but it is a simple way to do it. Having said that, not all maintenance templates are very good in categorisation: an issue that needs to be addressed. - 52 Pickup (talk) 08:15, 9 January 2008 (UTC)
This would actually be quite easy for a bot, particularly since, given the overwhelming number of cleanup templates on articles, you don't need a list that is updated hourly. I suggest posting at Wikipedia:Bot requests. Manually, you could pick a cleanup category and a subject/topic category, and all articles in both categories (including subcategories), using the category intersection tool. -- John Broughton (♫♫) 09:46, 9 January 2008 (UTC)
Many thanks for the speedy replies, everyone. I've brought it up on the bot requests page, would be great if it's possible! Thanks again, Seaserpent85 14:15, 9 January 2008 (UTC)

Broken "database locked" template

Does the template that displays when "The database has been automatically locked while the slave database servers catch up to the master" appear broken to anyone else? The lock image displays along with the above text but there is a bunch of wikicode displayed around it as well. I have no idea what template that is or who maintains it. --Spike Wilbury talk 14:43, 9 January 2008 (UTC)

I saw it too. The page is MediaWiki:Readonlytext. I'm not sure why its messing up, it hasn't been seriously edited in a long time. - Trevor MacInnis (Contribs) 15:35, 9 January 2008 (UTC)
For those who hadn't seen this yet. An example of how this looks is visible here --TheDJ (talkcontribs) 19:01, 9 January 2008 (UTC)
I've seen it too. Somehow it won't parse the table code. I could turn it into HTML... that should at least fix the problem for now. EdokterTalk 19:39, 9 January 2008 (UTC)

I filed a bugticket for this by request of brion: ticket 12567 --TheDJ (talkcontribs) 01:24, 10 January 2008 (UTC)

December 2005

There's some technical SNAFU with December 2005 making it uneditable by section. Can someone take a look? -- Kendrick7talk 19:59, 9 January 2008 (UTC)

OK, I found the work around, but this all is very confusing. -- Kendrick7talk 20:02, 9 January 2008 (UTC)

Developers' noticeboard (moved discussion)

this was moved from Wikipedia:Bureaucrats' noticeboard. EVula // talk // // 20:01, 9 January 2008 (UTC)

Please can we start a developers' noticeboard? I think there are many people who prefer the static nature of wiki talk pages rather than the horribly transient nature of IRC. OK, it probably should be in a MediaWiki domain rather than Wikipedia but we should note its existence and provide a link to it. -- RHaworth (Talk | contribs) 06:38, 9 January 2008 (UTC)

Wikipedia:Village Pump/Technical seems to fairly well watched by developers. There are also many people who aren't developers, or at least officially, that have the technical knowledge to answer inquiries. John Reaves 06:52, 9 January 2008 (UTC)

Bugzilla is what you're looking for. MER-C 12:02, 9 January 2008 (UTC)

As an aside, anything in the MediaWiki domain is uneditable by anyone other than an administrator, making it a poor choice for a noticeboard location. The MediaWiki namespace is also reserved for system-level pages (such as MediaWiki:Confirmdeletetext), and a discussion board doesn't qualify. EVula // talk // // 20:07, 9 January 2008 (UTC)

I think he meant the MediaWiki website. Mr.Z-man 21:16, 9 January 2008 (UTC)
Ah, upon re-reading the statement, I believe you're correct. EVula // talk // // 05:02, 10 January 2008 (UTC)

Protecting non-articles (moved discussion)

Discussion moved from Wikipedia:Bureaucrats' noticeboard. EVula // talk // // 20:03, 9 January 2008 (UTC)

Pending someone pointing me to the developers' noticeboard (see above), I am posting this here.

The recently introduced ability to protect titles even when there is no article is a very Good Thing. One small improvement is needed - a link to the log so we can see who did the protection. For example:

  • In the revision history of Awesomeness (which does exist as an article) protecting the title actually counts as an edit, as well as being available via the "View logs for this page" link, but for
  • Create my own page (which we do not want as an article) it is difficult to see who protected it. All we need is a "View logs for this page" link, please. (The "action=history" view does work - it is just that there is no link to it!)

-- RHaworth (Talk | contribs) 06:38, 9 January 2008 (UTC)

Clicking the "unprotect" tab will show you who protected it. John Reaves 06:50, 9 January 2008 (UTC)
Thank you, I had not noticed. But I would still like a "view logs" link. -- RHaworth (Talk | contribs) 06:57, 9 January 2008 (UTC)
Why is this thread here (WP:BN)? This belongs at Wikipedia:Village Pump (technical) (see above), or at the very least, WP:AN. I remember a Bureaucrat raising concerns over the overuse of this noticeboard for topics and discussions that sometimes doesn't have anything to do with Wikipedia:Bureaucrats. It appears this wasn't heeded. - Mtmelendez (Talk) 13:27, 9 January 2008 (UTC)
RHaworth was the one that inquired about where this type of question should go above (this was posted at the same time). As a side note, the concerns brought up by the last bureaucrat were unfounded and elitist. John Reaves 14:59, 9 January 2008 (UTC)
For the record, I don't see how a statement that non-crat-involved topics shouldn't go on the bureaucrat noticeboard is particularly elitist. Really, anything that doesn't pertain to current and proposed adminship promotion systems, renaming policies or methods, or bot bits doesn't need to be here (WP:BN). EVula // talk // // 19:59, 9 January 2008 (UTC)

Why is the Wikipedia search engine so bad?

Almost everyday of my life, I come across the wikipedia search engine and its all out incompetence. Every time, I have to resort to google to find the beginnings of a search that I eventually end at wikipedia. Can someone please explain to me why nothing has been done to fix this completely basic and essential function! Other people out there MUST agree that this is a serious and fixable issue. —Preceding unsigned comment added by Rmarti55 (talkcontribs) 03:46, 3 January 2008 (UTC)

Developers aren't paid, and there are many things on the to-do list. Hopefully it will get done soon, but there's not much one can do about it unless they have some programming skills. -- Ned Scott 03:52, 3 January 2008 (UTC)
Actually there is a very nice search engine. It is just that the more advanced features are disabled due to the amount of resources they use. In short, we are using our servers to keep the encyclopedia going and running an advanced search engine can be very resource intensive. 1 != 2 04:05, 3 January 2008 (UTC)
First, Wikipedia's task is not to search, but to allow editing pages. Second, we are used to excellent search engines, so any improvement that can be done here won't be enough. And third, Wikipedia has 100 servers, Google has 15-100x that amount. Just use Google to search Wikipedia ;-) -- ReyBrujo (talk) 04:11, 3 January 2008 (UTC)
What's wrong with the search engine? -- Tim Starling (talk) 06:02, 3 January 2008 (UTC)
Well for one thing, the summaries are poor: Compare Wikipedia search on Michelangelo to a Google Search. Google's rendition generally excludes templates and provides a better brief summary of context. In other cases the results are simply pretty lame: Wikipedia search for Mitochondria vs. Google Search. Google places mitochondrion first (as it should be), and lists several of the most major topics immediately after. Wikipedia puts Mitochondrion 15 places down and starts with a rather obscure fiction reference. Dragons flight (talk) 10:01, 3 January 2008 (UTC)
Yes, no contextual display of the search terms found in the articles. Also, I'm not sure about the searching syntax, and I'm not sure users less advanced than I could possibly be. MigFP (talk) 11:54, 3 January 2008 (UTC)
Our search engine is great for text search (after all, it's Lucene). But for semantic stuff, it's not. Google uses backlinks, which is part of the reason it can sense that mitochondria = mitichondion: people (e.g., bloggers) linking to http://en.wikipedia.org/wiki/Mitochondrion with the text "mitochondria" do the work for them. GracenotesT § 16:26, 3 January 2008 (UTC)
Thank you for your comments. It would be nice to have some of these things filed as feature requests in Bugzilla. The village pump is not the easiest thing for developers to search when they're looking for potential improvements. Here are the outstanding bugs against Lucene Search: [9].
We don't use Lucene's context highlighting, we use our own PHP code which dates back at least 5 years. Part of the problem with "mitochondria" is mis-stemming. We use Porter stemming, which according to [10] gives mitochondrion->mitochondrion, mitochondria->mitochondria and mitochondrial->mitochondri -- so only the precise word form "mitochondria" contributes to relevance. And we don't use backlinks or popularity-based ranking. For each of these features, software development is the bottleneck, we can probably afford the necessary servers. Google, after all, has to search the whole web, we only have to search a tiny corner of it. -- Tim Starling (talk) 00:39, 5 January 2008 (UTC)
  • That might be because the image is on Commons. Find an image that's on Wikipedia and try again, remembering that the default search excludes the image namespace. For example "AmiiStewartGreatest" finds the article that image is in. Include the image namespace, and the image comes up as well. I tried partial strings, and they didn't work, so that might be a weakness (but in some ways a good one). Carcharoth (talk) 16:29, 8 January 2008 (UTC)
  • I see. Thanks for explaining that. Given how very often images from Commons are used on en.wiki, I think the en.wiki search engine ought to be able to search automatically on en.wiki as well as on Commons. - Neparis (talk) 01:02, 11 January 2008 (UTC)
Try the new Google custom Wikipedia Search Engine. I tried to include a link to the results it gives for "mitochondria", but that URL is spamfilter blacklisted... --MikeVitale 14:44, 9 January 2008 (UTC)

Special:New pages only gives 50 pages

Special:New pages gives only fifty pages by default. I thought it used to give 500. Has something changed, or was I confusing it with something else? Is there a way I can set it to give 500 every time? Fg2 (talk) 01:57, 5 January 2008 (UTC)

You can bookmark the link http://en.wikipedia.org/w/index.php?title=Special:Newpages&limit=500 which will give you 500 pages. Tra (Talk) 02:05, 5 January 2008 (UTC)
Thanks! I'll do that. Fg2 (talk) 02:15, 5 January 2008 (UTC)
You can in fact get up to 5000. Prodego talk 03:01, 5 January 2008 (UTC)
That may be a bit more than I need right now, but I'll keep it in mind! Thanks Fg2 (talk) 21:49, 5 January 2008 (UTC)
5000? Sweet. Why don't they have an 'enter a number' feature, then? Asenine (talk)(contribs) 13:08, 10 January 2008 (UTC)

Searching

I was wondering if it's "possible" to have search results for more than one project.

In other words, if I were to enter global warming in the search box at the left, click on the search button, could the search function be set up to show results from (for example) Wikinews as well as Wikipedia?

If not, how difficult would that be to set up? - jc37 08:12, 8 January 2008 (UTC)

I would think that it would be somewhat difficult, but a handy feature might be to add links to the other projects that input the current search terms (ie: when looking for "global warming", the heading on our search page[11] shows similar links elsewhere[12][13][14]). Kinda nice middle ground. EVula // talk // // 16:57, 8 January 2008 (UTC)
Nod. But I thought that, since all the sister projects share the various servers, this should be possible? (After all, universal sign-in should be coming any day now : ) - jc37 00:21, 9 January 2008 (UTC)
File a Bugzilla request and it might just be (otherwise the devs probably won't know about it). MER-C 04:11, 9 January 2008 (UTC)
I would (and I may still), but my experience with Bugzilla so far has been to be pretty much ignored. I filed mediazilla:11499, months ago, and AFAICT nothing's come of it. But then, maybe I did something wrong in the proposing? (There were a lot of "options" for the uninitiated.) - jc37 11:26, 10 January 2008 (UTC)