Misplaced Pages

Revision as of 12:00, 28 August 2012

Jimbo welcomes your comments and updates – he has an open door policy.

This is Jimbo Wales's talk page, where you can send them messages and comments.

Put new text under old text. Click here to start a new topic. New to Misplaced Pages? Welcome! Learn to edit; get help. Assume good faith Be polite and avoid personal attacks Be welcoming to newcomers Seek dispute resolution if needed

Archives: Index, Index, A, B, C, D, E, F, G, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213, 214, 215, 216, 217, 218, 219, 220, 221, 222, 223, 224, 225, 226, 227, 228, 229, 230, 231, 232, 233, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 249, 250, 251, 252Auto-archiving period: 1 day

Archives

Index –index– October 28, 2001 – April 12, 2004 April 16, 2004 – August 23, 2004 August 28, 2004 – December 8, 2004 December 9, 2004 – March 23, 2005 March 23, 2005 – July 24, 2005 July 26, 2005 – September 9, 2005 September 9, 2005 – September 24, 2005 September – December 2005 January 2006 January – February 2006 February 2006 February 2006, cont. March 2006 April 2006 – late May 2006 May 24 – July 2006 July 2006 – August 2006 August 2006 Most of September 2006 Late September 2006 – Early November 2006 Most of November 2006 Late November 2006 – December 8, 2006 December 9, 2006 – Mid January 2007 From December 22, 2006 blanking Mid January 2007 – Mid February 2007 Mid February 2007– Feb 25, 2007 From March 2, 2007 blanking March 2–5, 2007 March 5–11, 2007 March 11 – April 3, 2007 April 2 – May 2, 2007 May 3 – June 7, 2007 June 9 – July 4, 2007 July 13 – August 17, 2007 August 17 – September 11, 2007 September 14 – October 7, 2007 October 28 – December 1, 2007 December 2 – December 16, 2007 December 15 – January 4, 2008 January 4 – January 30, 2008 January 30 – February 28, 2008 February 28 – March 11, 2008 March 9 – April 18, 2008 April 18 – May 30, 2008 May 30 – July 27, 2008 July 26 – October 4, 2008 October 4 – November 12, 2008 November 10 – December 10, 2008 December 5 – December 25, 2008 December 25 – January 16, 2009 January 15 – January 27, 2009 January 26 – February 10, 2009 February 8 – March 18, 2009 March 18 – May 6, 2009 May 5 – June 9, 2009 June 10 – July 11, 2009 July 12 – August 25, 2009 August 26 – October 3, 2009 October 4 – November 20, 2009 November 20 – December 21, 2009 December 22, 2009 – January 25, 2010 January 22, 2010 – February 22, 2010 February 22, 2010 – March 17, 2010 March 17, 2010 – April 9, 2010 April 11, 2010 – April 24, 2010 April 23, 2010 – May 12, 2010 May 10, 2010 – June 4, 2010 June 4, 2010 – June 26, 2010 June 26, 2010 – July 22, 2010 July 22, 2010 – August 23, 2010 August 4, 2010 – August 28, 2010 August 28, 2010 – September 14, 2010 September 13, 2010 – October 6, 2010 October 6, 2010 – November 8, 2010 November 8, 2010 – November 21, 2010 November 21, 2010 – December 6, 2010 December 6, 2010 – December 22, 2010 December 21, 2010 – January 17, 2011 January 16, 2011 – February 10, 2011 February 6, 2011 – March 19, 2011 March 16, 2011 – March 26, 2011 March 26, 2011 – April 22, 2011 April 22, 2011 – May 16, 2011 May 17, 2011 – June 9, 2011 June 10, 2011 – July 1, 2011 July 1, 2011 – July 15, 2011 July 16, 2011 – July 26, 2011 July 26, 2011 – August 15, 2011 August 14, 2011 – August 23, 2011 August 21, 2011 – August 31, 2011 August 30, 2011 – September 7, 2011 September 4, 2011 – September 25, 2011 September 26, 2011 – October 19, 2011 October 16, 2011 – November 5, 2011 November 4, 2011 – November 12, 2011 November 11, 2011 – November 24, 2011 November 24, 2011 – December 6, 2011 December 3, 2011 – December 14, 2011 December 10, 2011 – December 16, 2011 (RFC about the Stop Online Piracy Act) December 15, 2011 – December 30, 2011 December 27, 2011 – January 9, 2012 January 9, 2012 – January 18, 2012 January 17, 2012 – January 30, 2012 January 28, 2012 – February 12, 2012 February 7, 2012 – February 25, 2012 February 19, 2012 – March 4, 2012 February 27, 2012 – March 18, 2012 March 17, 2012 – April 2, 2012 March 31, 2012 – April 13, 2012 April 12, 2012 – April 18, 2012 April 18, 2012 – April 26, 2012 April 26, 2012 – May 6, 2012 May 2, 2012 – May 22, 2012 May 19, 2012 – June 9, 2012 June 4, 2012 – June 20, 2012 June 18, 2012 – June 28, 2012 June 26, 2012 – July 12, 2012 July 3, 2012 – July 22, 2012 July 19, 2012 – July 29, 2012 July 25, 2012 – August 9, 2012 August 7, 2012 – August 30, 2012 August 29, 2012 – September 13, 2012 September 13, 2012 – September 27, 2012 September 26, 2012 – October 5, 2012 September 24, 2012 – October 20, 2012 October 9, 2012 – October 31, 2012 October 29, 2012 – November 19, 2012 November 17, 2012 – December 5, 2012 December 2–15, 2012 December 15–28, 2012 December 24, 2012 – January 12, 2013 January 13–31, 2013 February 1–10, 2013 February 11–28, 2013 March 1–16, 2013 March 17–27, 2013 March 28 – April 6, 2013 April 6–19, 2013 April 19 – May 3, 2013 April 25 – May 8, 2013 May 6–19, 2013 May 18–30, 2013 May 27 – June 16, 2013 June 17–30, 2013 June 27 – July 30, 2013 July 31 – August 4, 2013 August 4–12, 2013 August 13–27, 2013 August 28 – September 4, 2013 September 5–10, 2013 September 11–22, 2013 September 24 – October 6, 2013 October 7–16, 2013 October 17–27, 2103 October 28 – November 6, 2013 November 7–19, 2013 November 20–27, 2013 November 28 – December 5, 2013 December 6–24, 2013 December 25–30, 2013 December 31, 2013 – January 14, 2014 January 15–24, 2014 January 25–29, 2014 January 30 – February 10, 2014 Archives not yet dated: 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 This manual archive index may be out of date. Future archives: 184 185 186

This page has archives. Sections older than 24 hours may be automatically archived by ClueBot III when more than 2 sections are present.

Templates 90% of slow articles

I have reconfirmed that use of large templates is slowing the reformatting, or edit-preview, of major articles, as over 90% of the total delay. For a long edit-preview, divide by 10 to find the non-template speed improvement: article "Israel" with edit-preview of 40 seconds would drop to 4-second edit-preview with no templates. While most of the articles are slowed when having over 70 {Cite_web}, {Cite_news}, {Cite_journal} or {Citation} templates, some articles are slowed more by other large templates. For example, article "Miami" (which has an edit-preview of 18 seconds average) has relatively few citation templates, so other templates account for over 60% of the total template-generated delay in "Miami" where Template:Miami_weatherbox runs almost 4 seconds, even though the typical huge top "{infobox settlement}" runs within only 1 second. I have noticed some other articles where citations were only half the problem, and other large templates delayed the edit-preview beyond the ideal 7-second response. I guess the general message is: all large templates have the potential to slow edit-preview of articles, not just citation templates. Meanwhile, plans to improve the speed of citation templates are still stuck in approval stages, for almost 2 months now. It took a long time to reach the slow edit-preview, and it has taken years to gain no approval for faster templates. -Wikid77 (talk) 17:50, 24 August 2012 (UTC)

• I'll confirm that I had at least a 20-second edit preview. But have you figured out how much of this is due to the size of the templates or the trouble of transcluding them, and how much due to assembling and formatting that very long list of "templates used in this preview" and the protection status of each, view source links, etc.? (I'm not sure if this is a valid test, but putting the whole article in "display:none" style reduced my display time by less than half.) Maybe that function could be relegated to some specialized Special: page and not included every time you preview an edit? Wnt (talk) 21:03, 24 August 2012 (UTC)
  • Not a valid test. The parser doesn't look ahead and say "well he doesn't want to see it, let's serve up nothing and go for a smoke break". All the same grinding and gnashing happens, and the same content is delivered to your browser. It is your own browser that interprets style directives (you could use local Javascript to turn display: on again), so what you are seeing is the time saved by not asking your own browser to render the page contents. Since all our readers generally do want to see the page, you've sort-of done a reverse (or perverse?) test that shows the time they do have to take. Franamax (talk) 23:43, 24 August 2012 (UTC)
    • D'oh, I shoulda known that! Definitely there in the source. Wnt (talk) 13:24, 25 August 2012 (UTC)
• Repeated large templates are slow, not total templates: The average of article "Outline of Florida" shows only a 1.5-second edit-preview (using button "Show preview"), despite the use of 41 templates, to format almost 20 scroll-screens of page text with 25 images, where almost everything is a wikilink, but still just 1.5 seconds for the edit-preview or reformat of the entire article, positioning 25 images/icons. That article "Outline of Florida" even uses the huge, but efficient Template:Citation/core, but only once with just one {Cite_web}, so the result was a mere 1.5 seconds to reformat. When large templates are used over 70 times per page, then the slow-down becomes more obvious. That is the reasoning behind new Template:Cite_quick, to format simple, common citations 10x-12x times faster than {cite_web} or {cite_news}, as in pop-culture articles which do not use Harvard referencing or author-to-title links. -Wikid77 (talk) 00:00, 25 August 2012 (UTC)
  • And this is something we have known (as a community) for years, yet we still have opposition to improvements, often from those who are in blissful ignorance, but sometimes from those who surely should know better. Rich Farmbrough, 01:44, 25 August 2012 (UTC).
    
    • Well, I am thinking since removing citation templates would be a violation of wp:CITEVAR without prior consensus, then changing short cites to use {cite_quick}, in the same format as Citation style 1 of {cite_web}, is the easy alternative. Any short citation could be expanded, for more parameters, by returning to {cite_news} or {cite_journal}, but meanwhile, the major pop-culture articles would reformat, or edit-preview, almost 4x times faster since {cite_quick} is 10x-12x times faster than the others. For popular medical articles, with complex citations, then {Fcite_journal} could be used, as 4-5x faster than {cite_journal} but still support Harvard referencing and other parameters. -Wikid77 02:28, 25 August 2012 (UTC)
      • I've previously commented on the spamming of navboxes to many articles; these in turn become very large navboxes. I suggested "a neutral bright line putting a limit on templates. I think that limiting navboxes to linking 50 articles or less is more than fair, and limiting them to 25 is reasonable." Unfortunately, interest in the problem was solely political. But I'll point out that Template:Miami-Dade County, Florida links to (and is linked by) 100 different communities in Miami-Dade County, and is an example of the sort of template I wanted to see replaced by a list article or Category. Wnt (talk) 13:38, 25 August 2012 (UTC)
    • Where has this nonsense about opposition to improvements come from? Far from there being opposition to improvements, there are quite a few people actively pushing Scribunto which seems to be a distinct improvement. The only significant opposition has been to the templates that Wikid77 made and started using, and that was precisely because they were not improvements. They broke quite a lot of stuff.

      Thanks to M. Starling, Scribunto is now on the test2 server, per the announcement in The Signpost a few days ago. I've set up a citation templates test page with a lot of templates in it there, that exercises a lot of our various citation, and cross-linkage, templates. Scribunto works to the extent that we can actually implement our citation templates in LUA. I have. Look how short Template:cite book is. All of the logic is in the LUA module (which as you can see by the test pages, incorporates a fair fraction of the functionality of the English Misplaced Pages's {{Citation/core}}) without having templates that transclude templates that transclude templates and closing braces coming out of one's ears. This is the improvement that people are working on, and I've yet to see it (sensibly) opposed.

      If you want something to compare, look at The Beatles#Citations and The Beatles#Sources here versus The Beatles#Citations and The Beatles#Sources on the test2 wiki, which is the same wikitext (albeit that the infobox and a few other templates are missing — I'm just about to do a LUA version of {{Allmusic}} for it.).

      Uncle G (talk) 21:14, 25 August 2012 (UTC)

      • Excellent work, thanks. That looks very promising (although I think we are still stuck with a mess when interpreting template arguments with problems handling = and | and more?). Has it been possible to estimate actual performance improvements yet? Do you mean T. Starling in above? Johnuniq (talk) 03:32, 26 August 2012 (UTC)
        • M. isn't an initial, and that's not really an interpretation problem.

          Scribunto doesn't change the wikitext syntax. So we'll still have to use 1= and so forth whenever there's an = character in a template argument, as now. What Scribunto changes is the potential implementation of templates, drastically, to one that works more speedily. In our current system, {{cite book}} expands into {{citation/core}}, which then has to be re-parsed and in turn expands into {{citation/make link}} and {{citation/identifier}}, which then have to be re-parsed and in turn expand into {{hide in print}} and {{only in print}}. In the Scribunto version of things, this is the entirety of Template:Cite book:

          <includeonly>{{#invoke:citation|citation|CitationClass=book}}</includeonly><noinclude>{{documentation}}</noinclude>
          

          That's it. There's one level of template expansion. The same goes for infoboxes. In our current scheme, {{infobox radar}} expands to {{infobox}} which after re-parsing in turn expands to other templates such as {{infobox/row}}. Again, the Scribunto equivalent has one level. The Scribunto equivalent of {{infobox country}} is also interesting for other reasons. The template is again one line long:

          <includeonly>{{#invoke:infobox_country|country}}</includeonly><noinclude>{{documentation}}</noinclude>
          

          The Lua module that the infobox implementor has to write is Module:Infobox country. That's not complete, but the missing parts are simply more of the same. See how simple in structure it is! Ask yourself whether you, assuming/pretending that you don't know Lua, could implement the rest of the fields in the infobox that I haven't implemented by just following the pattern of what's already there.

          As for performance improvements, see what I wrote on the Technical Village Pump. United States?action=purge loads in ~10 seconds on the test2 wiki (and it has an infobox template to process now). Compare that to United States?action=purge here. Or just open them up side by side in tabs and see which one completes first. ☺

          Somewhere in United States there's a template that in the current system expands to 28 levels deep ….

          Uncle G (talk) 11:16, 26 August 2012 (UTC)

          • Thanks again, although I have read some of the docs for Scribunto and Lua and understand that standard template syntax is used to invoke a Lua module (I was expressing disappointment that there is no magic way to clean that up due to backwards compatibility requirements). I just tried the purge and got 15 and 35 seconds for test2 and here, respectively. There are a lot of confounding factors such as the time required to purge a similar page with no templates, the hardware used for test2 and for here, and the current workload on each site, so we'll need to see how things work out when implemented here. I agree that a Lua module is much better than MediaWiki template voodoo. Johnuniq (talk) 12:04, 26 August 2012 (UTC)
            • Those and several others. This is why I pointed out on the Pump that these are not (yet) like-for-like comparisons. But one of the things that having this on the test2 wiki gives us is the ability to have like-for-like comparisons. Copy over a hundred or so articles that make heavy uses of citation, infobox, conversion, coördinate, and other templates; Scribble the appropriate templates; and compare and check. This is the whole testing thing that Wikid77 missed out on last time and seems to be missing out on this time. Uncle G (talk) 17:16, 26 August 2012 (UTC)
              • Well, that was the early phase and I did not "miss out" on testing, as I updated several articles in early July, which revisions now work fine. However, beware people falsely claiming "broken articles" when just a few links fail, or the errors were already in the articles, as broken articles before changing citations. Note that "India" has a few broken citations using "coauthors=" rather than last2/last3 for author-title links with {{sfn}}. Those were pre-existing errors in "India" not caused by the Lua script, nor Template:Fcite which works fine. -Wikid77 03:16, 28 August 2012 (UTC)
                • "Just a few links" being the entire shortened footnotes system, amongst others. More misleading terms here are your classifying template parameters, that you've just decided to not support, as "errors". Again, this is where you are completely wrongheaded. A drop in replacement has to be capable of being dropped in. That means supporting things like coauthors (and indeed coauthor — I've provided both in the Scribbled templates.) because uses of them exist. Ironically, but in demonstration of the point that you simply skip over the testing and checking parts, India has no citation that uses coauthors and hasn't had since July. Fortunately, one of the other articles copied over to the test2 wiki does.

                  To all of M. Wales' talk page watchers: If you have a favourite obscure citation template parameter that you want to ensure will continue to work with Scribbled citation templates, feel free to do as Allen3 (talk · contribs) has done and drop in a test case, copied from an actual current use in an article for preference, at the bottom of Citation templates test. The goal here is to support all of the existing functionality, and more besides, so that we have something ready to drop in when the time comes, as well as so that M. Starling has something akin to real use to profile and debug; rather than to arbitrarily throw away functionality and declare "fast citations". ☺

                  Uncle G (talk) 08:49, 28 August 2012 (UTC)

            • Here's an interesting datum that is a like-for-like test, because it's on the same wiki: I copied India over to the test2 wiki. It turns out that it uses {{citation}} directly, a lot. I hadn't Scribbled that template at the time. On the test2 wiki it took somewhere over 50 seconds for the page to display using the old non-Scribbled template. I Scribbled Template:Citation and that reduced the time for India to approximately 20 seconds. Uncle G (talk) 14:28, 27 August 2012 (UTC)
              • Interesting. I have some questions that I'll ask here rather than inject noise into the VPT discussion. I'm hoping you might ask Tim to provide some info on the 20 seconds: how much of that is Template:Citation, and is there a stand-out bottleneck apart from Citation? Is the Scribunto/Lua in use at test2 running at full speed (or is it running with some kind of debug/profiling code)? This is just my curiosity, but some info added at VPT (or a dedicated page somewhere?) would be interesting. Johnuniq (talk) 01:54, 28 August 2012 (UTC)
                • I have no special access to M. Starling, and if you want to ask xem something it's best done directly rather than through me. I'm one of the people writing Scribbled templates, and articles that use those templates. That puts me in the rôle of Complaining End User Who Wants To Do Unexpected Things And Have Everything Work Just Like In The Doco as far as M. Starling is concerned. ☺

                  But there's a figure for how much time is spent by Lua given in a comment in the generated HTML. I just re-purged India and looked at the figure. It reported 5.797s, out of just under 20s in total by my count, spent in Lua, and that's presumably the total for every Scribbled template, which includes {{citation}}, {{notelist}}, {{infobox country}}, and several others at this stage. (The remaining things to get a proper like-for-like comparison up and running are some missing fields in the country infobox, some CSS generation in citations, a few navboxes and ancillary non-Scribbled templates, and {{convert}}. Conversion might add significant time on. The missing country infobox fields probably won't. They're just a tedious exercise in decoding what multiply nested {{#if}} in the original are supposed to be doing. The non-Scribbled templates are fairly flat in the first place, which is why they don't need to be Scribbled.)

                  As for the other question: I'm guessing, since M. Starling has said that xe's going to profile things, that the test2 server has some form of profiling enabled.

                  Uncle G (talk) 08:49, 28 August 2012 (UTC)

              • One edit-preview of test2.wikipedia "India" ran only 14 seconds with the Lua script Module:Citation, which is comparable to using Template:Fcite, or 5-6x times faster than template-based {Citation} formatting. However, Module:Citation is still being revised, so it might become slower after it is expanded closer to {cite_web} or {citation} operation, with all 9 author names and "et al." formatting. -Wikid77 03:16, 28 August 2012 (UTC)
                • You're trying to grasp at straws, there. Module:Citation supported "all" 9 author names right from the start, when citation template support was first added to it, long before I even copied India across for testing. What you saw, supports "more than all". Uncle G (talk) 08:49, 28 August 2012 (UTC)

Speeding Miami to 6 seconds from 18

By a combination of faster techniques, the full article "Miami" can be improved to edit-preview in 6 seconds, rather than the 18-second reformat of recent months. The article "Miami" has been an interesting example, as a popular page viewed 4,500 times per day, because slow citations are only a fourth of the total reformat time, while other templates cause the majority of slowness. There are 4 techniques which apply here: preformatted template sections ("cached sections"), fewer bottom navboxes, fast-citation templates, and the "efficiency dividend" when running quicker pages:

• Preformatted template sections ("cached sections") - This was one of the tactics discussed last month, where a slow-template section could be run separately, and the generated results would be stored as if "cached" in a subpage. For "Miami", the cached section is {Miami_weatherbox/preformatted} which was developed from running the 4-second Template:Miami_weatherbox and copying the results into a custom-formatted subpage. Because the {weather_box} template is extremely complex (also listing humidex, snow, or humidity), rather than fight to make the complexity faster, the easy bypass is to cache the results into a subpage /preformatted, for instant inclusion in "Miami".
• Fewer navboxes: As typical for popular articles, "Miami" had over 9 bottom navboxes, where 7 were only remotely related to Miami, and are listed now by title-link only, not embedded as navboxes of "107" links each. That shaved about 2 seconds off the total reformat time.
• Fast-citation template {cite_quick}: Because the article was written using mostly {cite_web} or {cite_news} templates, then Template:Cite_quick could be used to format cites 10x-12x faster. This is very typical of pop-culture articles, such as films, celebrities, or news events, and {cite_quick} can shave another 4 seconds off the edit-preview time.
• Efficiency dividend: As noted last month, when a page is streamlined to run much faster, then the overall speed is further quickened by avoiding busy-server delays which affect the slower articles even more. In this case, the dividend is about 1.5 seconds, as if the original 18-second article could have displayed in 16.5 seconds in an "ideal world of no other users" which slow the servers during those dedicated 16.5 seconds.

All together, the total speed improvement of edit-preview becomes:

• 18 seconds - 4 (weatherbox) - 2 (navboxes) - 4 (cites) - 1.5 (dividend) = 6.5 seconds.

Because the busy servers (Monday-Friday) still affect even quick articles, then the 6.5-second edit-preview will occasionally stretch to 7 or 8 seconds, but reformat times of 10 seconds would be rare, and the original 18-second edit-preview would basically be long-gone from "Miami". The only catch is to gain approval to use {cite_quick} in major articles, as part of the wp:CS2 {Cite} template family, as intended. Then, repeat similar changes in other popular articles, and soon, many major large articles could be edited and previewed within a comfortable time period, within about 7 seconds. Q.E.D. -Wikid77 08:57, 25 August 2012 (UTC)

Excellent. I do prefer your "Smart citations" to a cite fork though. One more suggestion, embed the footnotes in the {{Footnotes}} and make that a subpage too. Rich Farmbrough, 13:57, 25 August 2012 (UTC).

• Cite_web/smart only 4x faster while Cite_quick 12x: The inescapable advantage of {cite_quick} is the 10x-12x faster speed compared to {cite_web/smart} as only 4x faster than {cite_web}. Perhaps there are other speed techniques which could be used to make those templates faster. Meanwhile, for {cite_web/smart}, I think I have used "every trick in the book" to keep it from slowing to only 3x faster. In fact, a {cite_journal/smart} is likely to be only 3x faster because {cite_journal} entries tend to use more parameters, and all of those will slow performance even more. Hence, the tactic for popular medical articles would be to use fast {Fcite_journal} as 5x-6x faster, because that would be almost 2x faster than a proposed {cite_journal/smart}. However, the potential to have a subpage as "article/footnotes/preformatted" is an excellent idea for some of the mega-articles in the top 500 most-viewed articles, such as "United States", to allow it to be edit-previewed as a shorter page, or reformatted with any image size, within a few seconds, rather than over 27 seconds with "Template include size too large". That footnotes tactic is so awesome, it will be like magic when editing "United States" again. -Wikid77 (talk) 17:46, 25 August 2012 (UTC)

With no change in the article wikitext at all, United States on the test2 wiki using Scribunto loads (with a cache purge of course) in 10 seconds. The infobox might slow it down a bit once added; but not, I suspect, by 13 seconds or more. And of course infobox templates are candidates for Scribunto, too. Uncle G (talk) 02:29, 26 August 2012 (UTC)

• With no change in the article wikitext except using {cite_quick}, "United States" reformats here in only 8 seconds, with the main infobox and all other templates still in use. Again, the primary advantage of {cite_quick} is the 10x-12x faster speed compared to {cite_web/smart} as only 4x faster than {cite_web}. However, {cite_web/smart} should be the upgrade version for {cite_web}, for use in the remainder of the 1.1 million cite-web articles which are not viewed, or edited, so often. -Wikid77 14:16, 26 August 2012 (UTC)
  • … which is a misleading way of saying with changes to the wikitext, because of all of the citation fields that your proposals intentionally don't support but that are used in that very article and that would outright lose displayed citation information if one didn't change the wikitext; such as doi= (used in 2 templates), quote= (used in 1 template), edition= (used in 2 templates), and month= (used in 16 templates) for starters. You did learn from your past mistakes and actually make sure that your purported go-faster system wasn't breaking things this time around, ne? Uncle G (talk) 17:00, 26 August 2012 (UTC)

Well, I did learn to beware people "crying wolf" when there are already problems with {cite_web}. Do not be fooled into thinking other templates are "not broken" when they also fail to process parameters. However, compare the following:

• {{cite web|title=See which parameters disappear | editor1 = Eddy | article=Parametercheck |Date=1 May 1944}}
• Cite_web result: Eddy (ed.). "See which parameters disappear". {{cite web}}: |article= ignored (help); Missing or empty |url= (help); Unknown parameter |Date= ignored (|date= suggested) (help)
• Fcite_web result: Template:Fcite web

At least {Fcite_web} notes the incorrect parameters. The goal is to edit-preview faster, to see the results faster, and adjust the parameters. Because {Fcite_web} is faster, it has more time to check for invalid parameters and warn the user. -Wikid77 08:59, 27 August 2012 (UTC)

Is anyone's time on Misplaced Pages, either as an editor or as a passerby reader, so constrained that shaving 12 seconds really makes a difference to anyone? Is this really the life we want to live in where we care about such inane things? Do we not have better ways of improving the encyclopedia instead of taking what works and trying to change it only for the sake of faster loading times? And at the expense of making the site user friendly and constructive.97.85.211.124 (talk) 00:50, 27 August 2012 (UTC)

Everyone's time: Here, the "shaving 12 seconds" is a small example, and many of the problems with major articles are 20-40 seconds longer to edit-preview, or view with imagesize set other than 220px. When the delay reaches 60 seconds, then the entire edit could be lost, although rare, due to wp:Wikimedia Foundation error. Plus, numerous people have been caring about such "inane things" for many decades, in far more detail with split-second timings; see article "Response time (technology)" for more there. When response time is faster, then a template has more time to check the user's text and suggest improvements, since the user will likely see the result within a few seconds, not 18, or 29, or 52 seconds later. -Wikid77 08:59, 27 Aug., revised 02:20, 28 August 2012 (UTC)

Javanese Misplaced Pages

The articles "Papat Limpad" and "Papat Limpad 2012" discuss efforts to revitalize Javanese Misplaced Pages.
—Wavelength (talk) 14:53, 27 August 2012 (UTC)

Editors can use Category:User jv to find Wikipedians who speak the Javanese language.
—Wavelength (talk) 03:05, 28 August 2012 (UTC)

Incivility

I am uninvolved in this issue, but i cam across Misplaced Pages:Administrators'_noticeboard/Incidents#Gross_incivility_from_two_editors, in particular the comment "fuck of and die you disgusting little heap of shit. Sociopathic scum like you". Thats not a death threat but I would think that is grossly uncivil to a congenial colloborative effort. To make matters worse, at the ANI thread the NPA "puke-brain" is reiterated. An admin seems to be working on it at User_talk:AndyTheGrump#Disappointing, but i thought this could be of interest to you. It would be something to look at in an effort to clean/improve the atmosphere. Disputes are quite normal, but that is not a dispute.Lihaas (talk) 04:15, 28 August 2012 (UTC)

I was going to refrain from further comment, but "Thats not a death threat but I would think that is grossly uncivil to a congenial colloborative effort"? Really? Of course it isn't a death threat - nobody has suggested it was. Why do you feel the need to point out that it isn't? And yes, it was grossly uncivil - it was intended to be. Again, nobody - including me - has suggested otherwise. You seem however to have conveniently omitted the context however, where "congenial colloborative effort" was clearly not the objective of the person I directed it at. Instead, we had a 'contributor' insisting that a mass murderer was not a terrorist despite the fact that not only had he been convicted of terrorism, but his own defence lawyer had described him as such (and no, he wasn't saying that Misplaced Pages shouldn't describe him as a terrorist, he was asserting that he wasn't one). Likewise, he was asserting that the mass murder of 69 individuals - mostly teenagers - wasn't 'an atrocity'. While I clearly shouldn't have blown my top like that, to pretend that this was some sort of 'content dispute' is a gross misrepresentation. Sadly though, yet again the petty amateur bureaucrats of Misplaced Pages seem intent on burying their heads in the sand, and pretending that all would be right in the world if we were more polite to each other. It doesn't work like that. Misplaced Pages has a serious (and almost certainly growing) problem with 'contributors' who exploit the 'rules' to their advantage while they spin articles to suit their own obnoxious agendas. In this case the 'contributor' even openly asserted on his own talk page that he "consider Misplaced Pages to be an intrinsically evil concept and a malevolent entity, a cancer on truth and on legitimate academic studies. Its concept of verifiability is the core of its evil. I am not here because I want to contribute to Misplaced Pages - I am here because I oppose everything Misplaced Pages stands for. A good writer does not bury truth because suitable sources are not immediately to hand, a good writer does not promote lies simply because sources exist that present those lies as if they were truth. But I recognise that few people have such high moral standards - or the courage or the knowledge to carry through with them". . No sign of "colloborative effort" there. No sign of anything but someone who doesn't give a damn about anything but his own toxic worldview. We should not be 'collaborating' with such individuals at all. I for one refuse to. So yeah, I said things I shouldn't have, and probably deserve to be blocked for it (frankly, I'm burnt out and don't really care that much either way), but don't try to kid yourselves that blocking one particularly obnoxious long-term troll and one foul-mouthed grump is going to solve the problem. While AN/I and the rest concerns itself with enforcing an entirely bogus 'civility' amongst contributors, a significant proportion of such 'contributors' are filling Misplaced Pages with hate filled bigotry concerning the rest of the world. As I said at AN/I, this is cognitive dissonance at its most obvious - a petty concern for internal civility combined with an abject refusal to recognise what is going on in our articles. Every article that touches on issues of ethnic or religious conflict for example has its collection of POV-pushers working behind the scenes to spin things their way, and almost all of them soon learn how to exploit a system that 'assumes good faith' even when confronted with self-evident malevolence. Obsessing about 'enforcing the rules' while ignoring how such rules are being exploited plays right into the hands of the bigots - and let's not forget that Anders Behring Breivik, the mass-murderer that was being whitewashed by this person we were supposed to be 'collaborating' with is a great fan of Misplaced Pages - I'm sure he found it of great use as a source for excuses and 'justifications' for his activities. It there is one thing that Misplaced Pages doesn't lack for example, it is deranged Islamophobic trolls, out to convince the world that a vast all-encompassing conspiracy is out to conquer the world, enslave our women, murder our children, and in all probability (if they are to be believed) drink our blood. Are we supposed to 'collaborate' with such individuals? I seem to recall that previous 'collaborations' with those pushing similar conspiracy theories didn't work out too well... AndyTheGrump (talk) 05:36, 28 August 2012 (UTC)

i fully agree with andy here. wiki is slowly but steadily being filled up with islamophobic editors, giving wikipedia a highly negative media publicity, especially due to the breivik trial (see e.g. )-- altetendekrabbe  05:54, 28 August 2012 (UTC)

You, Andy, and a number of others I could name here, are usually uncannily right in these situations about the character and motives of the person you harangue. (I don't know about this instance.) (1) Sometimes you're not. (2) That language and behavior is unnecessary, so on those rare occasions when you're wrong it is an unnecessary gross mistreatment of another editor. (3) Whether you're right or wrong, it always consumes large amounts of other people's time. Show some consideration to the rest of us and take it somewhere other than the article talk page. --175.38.128.99 (talk) 05:57, 28 August 2012 (UTC)

I'm not involved in this conflict, just want to point out that "atrocity" does strike me as a POV term. As for "he wasn't saying that Misplaced Pages shouldn't describe him as a terrorist, he was asserting that he wasn't one", I don't see any evidence that that's the case. His comments on the Anders Breivik talk page were all focused on the use of the term in the article, not in general. I disagree with him on that point - reliable sources call him a terrorist, so it's fine - but rather than debate the issue, it looks like the editors in question quickly resorted to nasty personal attacks. Whatever Meowy's motivations were, telling him to "fuck off and die", among other things, is not in the least justifiable. DoctorKubla (talk) 06:11, 28 August 2012 (UTC)

The guy was an obvious troll, with a block log the length of your arm, and was trying to provoke - unfortunately he provoked two good Wikipedians (who really do care about the quality of the encyclopedia and how it progresses) into using injudicious words. I'm a great supporter of civility, but the context and circumstances matter too - we're not all robotic automatons who can keep emotion completely out of our dealing with others, especially not when they are trolls and are picking on a very sensitive topic to start a fight. -- Boing! said Zebedee (talk) 09:38, 28 August 2012 (UTC)

• Arguing that the murder of 69 people wasn't an atrocity is extreme. Note the editor's own comments here: : " consider Misplaced Pages to be an intrinsically evil concept and a malevolent entity, a cancer on truth and on legitimate academic studies. Its concept of verifiability is the core of its evil. I am not here because I want to contribute to Misplaced Pages - I am here because I oppose everything Misplaced Pages stands for." The incivility was uncalled for and unnecessarily inflamed the situation; I would have suggested filing at ANI instead. IRWolfie- (talk) 09:41, 28 August 2012 (UTC)

• • In the old days, I would have just personally blocked the troll on sight, and that would have been the end of that. One of the things that makes wikis work is precisely the ability of the community to tell people to knock off the nonsense or get blocked.
  • If you go back to the disastrous culture of unmoderated Usenet groups, you can see what happens if it is too difficult to block trolls from participation. What happens is that good people reach the end of their good humor and lash out. The social environment degrades to people screaming at each other and it becomes quite hard to tell the good people from the bad. If someone says that they "consider Misplaced Pages to be an intrinsically evil concept" then the solution is not to get emotional and lash out at them in anger, but to realize that telling them to fuck off is not nearly as satisfying as maintaining a good sense of humor while making them fuck off (with a permanent ban). We have better things to do!--Jimbo Wales (talk) 12:00, 28 August 2012 (UTC)

Misplaced Pages Pillars

Mr. Wales, I like Misplaced Pages a lot. I like depth and breadth of it. I can easily get lost reading about moths, and trees, and dinosaurs, and rare earth metals, and before I know it, I find I've wasted half of a Saturday morning. From one side, it looks really amazing and impressive.

But as an editor, I have to say, it feels like so much of a battle, so much of the time. While civility is a common practice in polite society, it seems to quickly become scarce in Misplaced Pages between our editors. And while I don't like to paint with broad brushes, the Administrator's Noticeboard probably running right behind Mos Eisley as a wretched hive of incivility. I like to imagine that our Pillars actually mean something here. And I would like to think that among all the editors we have, our Administrators would take our Pillars in hand with the utmost respect and professionalism. Unfortunately, you are more likely to get attacked by the regulars at Administrator's Noticeboard than helped. If you're new and don't understand things, you'll get attacked for being new and stupid. If you try to stick up for other people, you'll get attacked for that. Well, I won't belabor the point, but you get attacked mostly for just not agreeing wholeheartedly with whatever the mind of the mob is for that day.

Is WP:Civility as a Pillar a lost cause? -- Avanu (talk) 08:35, 28 August 2012 (UTC)

The current case cannot be defended (the incivility was extreme), but as has been pointed out on the many other occasions you have raised similar matters, the actual problem that the community cannot deal with is the civil pushing of POV and FRINGE views that drive good editors to incivility. Yes, the incivility needs to be dealt with, but it is a symptom of the underlying problem (the inability to deal with POV/FRINGE pushing because passers-by derail noticeboard discussions with links to AGF/BITE/CIVIL).

Problem editors need a clear message at noticeboards: we're sorry someone was rude, but please change your ways if you plan to do more editing because what has happened so far is not acceptable (link to policy goes here). Johnuniq (talk) 09:26, 28 August 2012 (UTC)

While I agree 100% with your comment, Johnuniq, I specifically raised a concern here because AN/I is broken. It isn't *totally* broken, and in fact a good 70% of incidents pass through it with a good outcome. But WAYYYY too often, people are snarky, dickish, and patronizing, rather than professional and courteous. I realize that these are volunteers, but they are unmanaged and uncoordinated. So while you have moments of splendor and touching service, you don't have consistency. And in light of the ideal of Civility, being dickish a large part of the time, and an angel the rest, doesn't make for that great of a saint. -- Avanu (talk) 09:55, 28 August 2012 (UTC)

{e/c}

I disagree . I've seen lots of gratuitous rudeness and smart-mouthing and arrogance. Civil (well-mannered) people don't often allow disagreements to turn them into sarcastic cussers. A better message could be, "We're sorry someone was pushing their POV, but please respond civilly." Yopienso (talk) 09:59, 28 August 2012 (UTC)

Fine. But the actual problem is that Misplaced Pages is now the number one forum for promoting ideas and products, and the old bazaar model that all problems will be solved by the inexhaustible supply of good editors is in danger of being submerged. There is an evolutionary arms race in play: after normal procedures select out those who can quickly be identified as bad, we are left with a group that is resistant to standard cleanup (the civil POV pushers). If we focus on CIVIL, we will all be very happy and nice, but the actual problem will deteriorate, and the encyclopedia will suffer.

My comment above is asserting that if there is an ANI report about a problem user, the discussion in that section needs to focus on the actual problem. If someone is uncivil to the problem user, by all means raise it with them on their talk page, or start a new section at ANI or whatever. But if there is a problem user being discussed, it is not helpful to derail the discussion by raising other issues. There have been many examples where problem users fail to get the message because they regard any objection to someone who was opposing them as support for their problematic behavior (an invalid conclusion, but they cling to any hope). I am not talking about the case currently on everyone's mind, as that is over-the-top. Johnuniq (talk) 11:17, 28 August 2012 (UTC)

Paid advocacy

"The tale of GettingBookReviews.com, which commissioned 4,531 reviews in its brief existence, is a story of a vast but hidden corner of the Internet, where Potemkin villages bursting with ardor arise overnight."

"Rutherford would review a book for $99. But some clients wanted a chorus proclaiming their excellence. So, for $499, he would do 20 online reviews. A few people needed a whole orchestra. For $999, he would do 50. Not just any kind of review will do. They have to be somewhere between enthusiastic and ecstatic. Before he knew it, he was taking in $28,000 a month."

http://finance.yahoo.com/news/best-book-reviews-money-buy-131408538.html?page=all

-- Avanu (talk) 11:45, 28 August 2012 (UTC)