User talk:Dispenser/Archive 3

FYI

[1], Happy new year, --Flominator (talk) 19:23, 2 January 2009 (UTC)

Coordinates log (Warning: UTF8params)

I'm not quite sure if anything should be done about a series of entries in the coord-enwiki log, e.g.:

Samples

• Bocock_Peak_Provincial_Park (with 'params=55_51_N_122_56_W_\xe2\x80\x8bregion:CA')
• Beatton_River_Provincial_Park
• Bear_Glacier_Provincial_Park
• Butler_Ridge_Provincial_Park

Their common point seems to be that a (in these cases useless) name precedes the coordinates, e.g. {{coord|name=Bocock Peak Provincial Park|55|51|N|122|56|W|​region:CA|display=title}}. -- User:Docu

There is another possibly related problem with UTF8. I mentioned it at Template_talk:Coord#Parser_function_errors_(Antelope_Valley_College). -- User:Docu

Para answered it in the meantime. -- User:Docu

Readability counts the page size incorrectly?

I was just adding references to Keith Miller in preparation for a FAC run but got suspicious when the "prose size" ballooned by about 5k from 85k to 90k, which made me suspicious, then I checked with Dr pda's script and it said that readable prose size was only 78k for Miller. It seems strange because I checked a few of my FA/GAs and in all the other cases the size was identical to within 1-2k. YellowMonkey (click here to vote for world cycling's #1 model!) 00:27, 4 February 2009 (UTC)

The readability program processes HTML, but the tool gets the wikitext and needs to parse that into HTML. The parse is a crude parser written in mostly regular expressions and it needs to be preprocessed to remove non-readable prose. Additionally, the readability approximations are poor due to the syllable counting algorithm. As I do not have much interest in this field; I have since ask again to have the tool delisted. — Dispenser 05:06, 14 March 2009 (UTC)

locateCoord.py

The below should allow to specify a distance from the first coordinates to center the selection around it. -- User:Docu

locateCoord.py

(source removed: solved)

Thanks for the implementation. Works well for retrieving coordinates.

BTW region codes with CH-NW, DE-BW and MW don't appear in the results (sample: Stansstad, Stuttgart or Dedza District). The "W" as the second letter of the country/subdivision code seems to throw it off. "US-WA" works. -- User:Docu

It works now. Great. Thank you. BTW I asked for pagenamee to be added to coor URL for it/de/es. -- User:Docu

I didn't have much luck for it/de/es. Anyways, the tool is broken right now, but I see you are working on it. I will try to improve the formula for the range calculation. Distance to the point of origin and links to the different articles in the form [[Place with wrong coordinates]] would be helpful as well. -- User:Docu

it/de/es are ok now. BTW ghel could use a fix at lines 281+288 for coordinates with negative N/E values involving min/sec (SW2b, SW3b on [2], NE3/NE3b is an input error). Further on, for the "range excessed" test to work, it would need float(). -- User:Docu

Something like the below might fix it, the test cases are ok now except for NE1b. BTW could you run the batches an hour earlier? I'm wondering what the results from iw would be now. -- User:Docu

d/m/s still needs a fix (tests SW3b, NE3b below). Negative values South and West should probably be an error, but I'm not sure how frequent they are. We could log them and then decide.

Not sure how "heading:" is to work, but it floats the log for commons. The "zero assumed" warning seems frequent in some languages, possibly because they use templates differently, maybe it shouldn't be activated there. For the range check, ideally 60 min/60 seconds would be of type "warning", everything else "error". Anyways, thanks for help and time for my previous requests. -- User:Docu

fix

Tests at [3] (except NE1b) and [4]

(source removed, partially solved)

The other day, I installed the mw GIS extension. A feature I liked, was the direct link to the list of places nearby on top of the map source page.

By moving up the links to locatecoord.py, this would be fairly easy to implement with GeoTemplate. I was wondering what you think of that. I'm not quite sure how frequently the mapsource and locatecoord pages are already used and how much additional traffic locatecoord could bear.

The GIS extension includes the distance and direction from the point of origin. I'm not sure if you want to add the distance on locateCoord.py results as well. [5] has a function to calculate the distance. -- User:Docu

Should be no more load than the region lookup, but I am going to be quite busy for the next 2-3 weeks. So I wont be able to implement anythings much less than explain. — Dispenser 03:38, 11 May 2009 (UTC)

Ok, good luck with the examens. I will try to answer the easier questions in the meantime. -- User:Docu

PDFbot

Hi Dispenser, I see from the archived discussion that there was an issue with PDFbot (talk · contribs) at the end of last year. Are there any plans to have it restart? I dread the thought of trying to fill in the sizes by hand. Thanks, BanyanTree 11:12, 25 February 2009 (UTC)

It was a miss understanding of the citation template. I do not believe that they've solved the double url problem yet. I'll be working in the meantime on improving the commonfixes. — Dispenser 05:06, 14 March 2009 (UTC)

Changing File: to Image:

Checklinks is changing links with the file namespace ([[File:Example.png]]) to the image namespace. Is this a bug or is it intended this way? 71.164.134.228 (talk) 06:04, 13 March 2009 (UTC)

The chatter in Bug 44 (Rename the "Image" namespace to "File") indicates a preference toward emphasizing the image aspects of the namespace and it is also nice to have a standard way of differentiating file type; not by their extension. It will only change from File: to Image: for images (otherway around for PDF, OGG, OGV, ...) when atleast one image is uses that notation. — Dispenser 05:06, 14 March 2009 (UTC)

I think you should include a mention of this at the bot's page, as it seems to be a common issue. I came here to ask about this as well. -RunningOnBrains^(talk) 04:48, 13 December 2009 (UTC)

User:Neurolysis/Toolbox

Hi, I have changed this back - it was meant to be like that specifically because of links like the one to NI. Thanks again, — neuro^{(talk)(review)} 04:39, 17 March 2009 (UTC)

You're invited!

{{Wikipedia:Meetup/NYC/Invite/March 2009}}
This has been an automated delivery by BrownBot (talk) 19:11, 21 March 2009 (UTC)

simple

hi dispenser, love the app. ever consider making it into an automated simple.wikipedia input bot? simple wikipedia really needs some help. i've played around with this idea a fair amount, and even automated simplification http://spencerwaterbed.com/soft/simple/ wanna collaberate? Spencerk (talk) 18:43, 21 March 2009 (UTC)

I would love to work on more language analysis projects and you have an interesting idea with a “English to Simple English machine translation”. However, I’m been short on time and I would really like to finish and polish the projects I’ve already started (this includes readability.py). And the scope of the project is larger than I can take on in the foreseeable future, sorry. — Dispenser 19:33, 28 March 2009 (UTC)

Reflinks

I ran Reflinks on an article but it died. It seems to die on articles with over 150 references; can you confirm this? Gary King (talk) 01:20, 27 March 2009 (UTC)

Also, the script shouldn't be changing File: to Image:. Gary King (talk) 02:50, 27 March 2009 (UTC)

See #Changing File: to Image:, above. — Dispenser 16:43, 7 April 2009 (UTC)

The script dies if it tries to crawl http://www.faceoff.com/story.html?id=a04cce42-f515-4e06-b99f-d8ed9ee21ca3. I think it might be the apostrophe in the page's title that's causing the problem. Gary King (talk) 19:25, 2 April 2009 (UTC)

I have run both webreflinks and reflinks on pages with 200+ references in the past without any issues. An explanation could be with the new web server software killing the thread if it was taking to long (unlikely). Over the weekend I added traceback auto mailer, so if there any exceptions in the future I’ll know about it.

 Fixed The tool die not because of the apostrophe (which is invalid UTF-8 character, btw) but because hex() did not do what I thought it did (convert Hex into a number). — Dispenser 16:43, 7 April 2009 (UTC)

Readability Tool

Hi,

I'm a computational linguistics student writing my thesis on stylistics in Wikipedia - and so am very excited to come across your readability statistics tool (thankfully also written in my favourite language :)

I'm wondering, though - where does one get the "syllable" and "wikipedia" python packages? Also, are there any other versions of the tool (apart from tools:~dispenser/resources/sources/readability2.0.py) available?

Cheers! Aengus —Preceding unsigned comment added by Ventolin (talk • contribs) 13:36, 4 April 2009 (UTC)

Unfortunately, the tool isn’t actually any good, sentence and syllable detection are way off. You are better off using the GNU Style and Diction programs which is going to the basis of a future version. If you really want to use my sources; 1.4 is the latest "stable", 1.5 is the test branch, 2.0 is a scraped branch, get syllable.py and pywikipedia.py is the same directory, however you likely want to use the regular m:pywikipedia.

I do not posses the interest or much time to develop this area for Wikipedia, but am willing to assist anybody else who wishes. — Dispenser 16:43, 7 April 2009 (UTC)

Votic language

Is this what you call « formatting » ?

Budelberger (   ) 22:46, 17 April 2009 (UTC).

So I forgot to type in the pipe character, I'm only human. — Dispenser 14:38, 16 June 2009 (UTC)

You're invited...

{{Wikipedia:Meetup/NYC/Invite/May_2009}}
This has been an automated delivery by BrownBot (talk) 21:50, 3 May 2009 (UTC)

Bot error

Kindly note this little glitch. ... Kenosis (talk) 21:38, 4 May 2009 (UTC)

I finally learned how to tackle that special case: (?!([^<]|<(?!ref))*</ref>). Thanks for the report. — Dispenser 14:38, 16 June 2009 (UTC)

New reports

Wheeeee. :D --MZMcBride (talk) 19:24, 5 May 2009 (UTC)

False positive?

What happened here? Is there a problem with using PDFlink in a cite template because it's redundant with format=PDF or something? If so, shouldn't the bot add in format=PDF instead of simply removing PDFlink? Also, what happened with this "agancy" thing down here, which has the effect of removing the agency from the footnote, since it's an unrecognised parameter? 79.64.170.147 (talk) 21:09, 6 May 2009 (UTC)

There typically always is a problem with mixing a formatting template into a citation template. The way it is done there is hack; using the fact the {{citation}} is a multipurpose template. If it was done using {{cite web}} it would error out. Additionally, many tools and parsing scripts except that the title parameter does not contain a URL and that the url parameter contain just a URL. It was not adding |format=PDF since it was a special run for all the incorrectly formatted template from WP:NRHP, I'll look into changing the for the future.

An old bug User talk:Dispenser/Reflinks#Reflinks - 'agancy', I should upgrade the version of commonfixes. — Dispenser 14:38, 16 June 2009 (UTC)

geo datas

Hello, I will use your geocoordinates from database "u_dispenser_p" now for WikiProjekt Wikipedia-World. So I play a little bit with your datas and get from you 677.000 coordinates for the main language where we had before 264.000 coordinates. Thats a big advancement and makes not so much work than to try to read the dumps with all stupid templates in it what we do in the past. I believe you use the links to geohack in the external link table. I write a tool to merge all language versions. So my first question is how actual are your datas? What do you do with this datas? Later I will have some special questions. --Kolossos (talk) 21:44, 9 May 2009 (UTC)

Update is daily (overview). It's used for locateCoord.py linked there as well, the daily error reports and the interwiki import report at tools:~dispenser/resources/logs/. I think Dschwen uses it for WMA. Dispenser will probably give you a more detailed answer. -- User:Docu

Here is my next question: Why you have only one de:Pesterwitz in the database? It is an article with many coordinats and use the parameter "name". On other articles like de:Liste der Namensgeber von Gebäuden der Technischen Universität Dresden it works fine. Do you have a blacklist or something like that? Where is the source of your script? --Kolossos (talk) 14:26, 10 May 2009 (UTC)

The source is in two files in the directory linked on the interface of locateCoord, one is ghel.py. The answer to your question is in its line 611: in the article "de:Pesterwitz" both coordinates/URLs are the same. -- User:Docu

That number is a bit inflated since I do not have a filtering program to remove nearly overlapping coordinate. More or less my todo list:

• Develop an API capable of writing out in HTML, JSON, serialized PHP, and XML.
• Language independent article ranking table (length, incoming links, interwiki links)
• Reset primary bit for multiple primary coordinates form the same article
• WikiMiniAltas/OSM data integration.
• Reimplemented features into GeoHack.
• Documentation, source code should be documented so a novice could understand it.
• Writing what silly things Docu want this week. ;-P

It all programmed in python; however, I would like make the data accessible to almost any programming paradigm. — Dispenser 14:38, 16 June 2009 (UTC)

Here some additional things for your to-do list ;-):

• "Develop an API capable of writing out in HTML, JSON, serialized PHP, and XML."

• +KML +OSM

• "Language independent article ranking table (length, incoming links, interwiki links)"

• +ranking with view statistic which I publish at database u_kolossos_wp_logs_p on toolserver (I get it from User:Melancholie.)

• "WikiMiniAltas/OSM data integration."

• Bring the POIs to tiles on different zoom levels so that we can handle very high load without to kill the database. With the project of integrate OSM-maps into Wikipedia articles we should have a layer of Wikipedia articles there. The load will be much higher as now. The performances of my script is not bad but perhaps not good enough for this case. (The priority is in the moment on openlayers, not on wikiminiatlas.)
• If you know the old google earth wikipedia layer of me and Stefan Kühn you know that it was very colored with a lot of different styles generated by different categories. It looks very good in my eyes. If you look now in my google earth layer (i use your datas now) too many things have the monotone icon of "landmark". So my greatest wish would be if you could also define different types of articles for the map with the help of the category system. Stefan and I would try to help you with your old definition system.

What do you think about it? --Kolossos (talk) 20:40, 22 June 2009 (UTC)

WikiBlame

Hi Dispenser, I just moved WikiBlame to SourceForge for easier development and translation. I currently didn't migrate your requests from de:Benutzer Diskussion:Flominator/WikiBlame to the tracker there. Can you maybe copy the ones that are stil relevant into new tracker reports? Do you maybe want to have subversion access (but be warned of the code)? --Flominator (talk) 12:43, 23 May 2009 (UTC)

I am not perl programmer so the subversion thing wouldn't be any good, but I'll look into re-evaluating those and making them clearer when I have time. — Dispenser 14:38, 16 June 2009 (UTC)

It's php, actually. Feel free to comment on the bug tracker. Regards, --Flominator (talk) 16:03, 21 June 2009 (UTC)

GeoTemplate SPER

You seem to be active on GeoTemplate. Could you comment on the SPER for adding elected.biz? Thanks, Celestra (talk) 14:30, 19 June 2009 (UTC)

I do not comment on individual services, sorry. — Dispenser 09:46, 21 June 2009 (UTC)

PDFbot changing wikinotation to HTML

That is, changing <br> to <br />, a useless edit.--BillFlis (talk) 18:06, 22 June 2009 (UTC)

See Wikipedia:Village pump (technical)/Archive 60#Changing <br> to <br />. — Dispenser 23:21, 22 June 2009 (UTC)

Checklinks not working correctly

I was using checklinks and noticed that the tool isn't fixing links correctly. For instance, I tried to change the following links on Video game console:

8	accessdate=2008-07-21 date=2008-07-21 publisher=Eurogamer	Hirai wants PS3 to beat PS2 [eurogamer.net]	301	Redirect does not contain ".,?&"
	publisher=Gameplan author=Forster, Winnie	The Encyclopedia of Game Machines - Consoles, handheld & home computers 1972-2005 [gameplan-books.com]	302	Changes domain and changes path
9	accessdate=2008-02-24 date=2008-02-19 publisher=Cnet	HD DVD: Just another brick in the wall of defunct formats [news.com]	301	Changes domain
17	accessdate=2007-06-09 date=January 29, 2004	Nintendo DS Details Explosion - Screen, Battery, GBA Compatibility and More [spong.com]	110	Connection timeout

I wanted to update the URLs on the first three and add in an archive link for the last one. When I save changes, archiveurl and archivedate tags are added into the first two sources with the URL as the paramter. Nothing is done for the second two. I also had similar problems with my first saved edit on History of video games and the draft of Downloadable content. Thanks. —Ost (talk) 16:50, 13 July 2009 (UTC)

Thanks for the reply! I'm glad to know you got the tool working again. I picked this page as an example of the symptom with a small amount of links, but I tend to be more situational about changing working http redirects. On that subject, it would be great if you could have a function that allows you to type the new URL and the archive URL; it would be helpful for when there is an archive of a page, but the page has been moved to a new URL without a redirect. Currently, I have the tool add the archiveurl and then manually update the url. It would also be useful if you could search WebCite along with the Wayback for archives. Thanks again. —Ost (talk) 18:05, 13 July 2009 (UTC)

I'm still getting an error. It looks like there is a typo—exapnd instead of expand—in the main() of mergeChanges.py at line 176. Sorry if you've already fixed this. —Ost (talk) 18:17, 13 July 2009 (UTC)

It looks like you fixed it. Thanks again. —Ost (talk) 18:18, 13 July 2009 (UTC)Sorry for another message. It looks like I was wrong. It works fine unless you try to add an archive link, which then causes the "expand" error. —Ost (talk) 18:21, 13 July 2009 (UTC)

Your coordinates log

Hi Dispenser, I hope this is the correct place for my message. I just took a look at your log (tools:~dispenser/view/File_viewer#log:coord-dewiki.log) and almost all errors are because of InvalidRegion. Please note, that the following region codes are used in German Wikipedia: XA, XI, XN, XO, XP, XS (see de:Vorlage:Coordinate#Ozeane). Maybe you can add those region codes as valid. Thanks --тнояsтеn ⇔ 11:04, 19 July 2009 (UTC)

I have added these code plus many of the transitional code into the set. This feature is still in development. — Dispenser 13:16, 19 July 2009 (UTC)

Thanks --тнояsтеn ⇔ 13:39, 19 July 2009 (UTC)

I have one question about this logfile, too. What happened to it? It is so huge that my browser crashes when I try to open it. I just saw before that de:Ägyptisches Museum Berlin is listed but there is no error in the coordinate. Thanks for you answer --92.229.118.30 (talk) 19:50, 19 July 2009 (UTC)

I inverted the conditions for an error check causing ~180,000 coordinates to report an error. Sorry, about huge file size crashing your browser. I've added warning for large files to the File viewer. — Dispenser 23:39, 19 July 2009 (UTC)

Thank you for fixing this. By the way: very helpful tool for geocoding! --92.229.83.157 (talk) 15:54, 20 July 2009 (UTC)

Alt text viewer

Thanks for letting us know of the new alt text viewer tool. I tried it out and find it quite helpful. I'd like to mention this on WP:ALT, and would like more info about it to do that. Can you please let me know:

• Where's the source code? I briefly looked at toolserver.org and couldn't find it. If I could read that I could answer many questions myself, without having to waste your time.
• I visited Quark and have some thoughts and suggestions:

• It didn't report alt text for <math> images; can you please add this to your list of requested features?
• The phrase "Place holder text" can be removed, surely.
• The phrase:

'These are non caption images, they may not need alt descriptions if they are simply decorative'

might be better worded as:

'The following images do not have captions. If an image is purely decorative it should be marked with "|link=" so that it does not function as a link to the corresponding image page. If the image is not purely decorative, it needs alt text.'

• It might be helpful to adjust the first (image) column to the right, rather than to the left, so that images are close to the alt text that describes them.

• I visited The Lucy poems and have another suggestion:

• The phrase "These are non caption images..." should be removed if there are no non-caption images.

• I visited List of Olympic medalists in athletics (men) and have another suggestion:

• If there are duplicates (i.e., the same image with the same alt text or lack of alt text), then list each duplicate just once.

• I visited WP:ALT and the tool crashed with a backtrace; you might want to look into that. Hmm, but now I see "The Lucy poems" didn't work either. I expect that you're hacking away at this now, so I'll stop for now.

Eubulides (talk) 20:41, 19 July 2009 (UTC)

The breakage was me renaming all extensions of the interface elements. Anyway, I've incorporated some of your wording suggestions into User:Dispenser/Altviewer/messages. I know the math markup is missing, but I would like the community to concentrate on images first. On the issue of the layout, I would actually like them to be further apart so reviewers aren't looking at the images while reading the alt text. The tool is very simple and scrapes every linked image off the page. It could cause more confusion (I have 14 images but only 8 are showing) if I were to eliminate duplicates. — Dispenser 22:58, 19 July 2009 (UTC)

Thanks for the explanation. I added an external link from WP:ALT to the Altviewer. I have some more feedback below. Eubulides (talk) 18:21, 20 July 2009 (UTC)

A bit more feedback

The tool is really useful and I hope you don't mind if a satisfied user has a few more suggestions. I visited http://toolserver.org/~dispenser/cgi-bin/altviewer.py?page=Inauguration_of_Barack_Obama and came up with these ideas:

• It'd be helpful if there was a convenience link to the original article, in this case to Inauguration of Barack Obama.
• The message "The following images do not have captions." turns out to be not quite right. In the following list are File:US President Barack Obama taking his Oath of Office - 2009Jan20.jpg, File:Inaugural invitation 2009.jpg, and File:44 CongInvitation.jpg, File:Barack and Michelle Obama at the Home States Ball.jpg, and File:20090120 Jill and Joe Biden at Homestates Ball.JPG. However, all of these images do have captions that are visible to the user; it's just that the Altviewer (understandably) can't see the captions because they are not generated using the image syntax. Perhaps the message should be reworded to something like this? "The following images do not use thumbnail syntax."

Thanks again. Eubulides (talk) 18:21, 20 July 2009 (UTC)

The script does not pickup the image File:LesterEliza.png in The Day the Violence Died. Gary King (talk) 18:48, 20 July 2009 (UTC)

Fixed the bug on The Day the Violence Died (Regular expression wasn't matching newlines). Again, you can modify the messages at User:Dispenser/Altviewer/messages. I've added some statistics and a convenience link. — Dispenser 23:21, 20 July 2009 (UTC)

"I am stumped on what alt text is/does/checks for", and that is the reason why I stopped writing documentation. Of course it probably would have helped if the link to WP:ALT was point to Wikipedia servers... — Dispenser 22:04, 27 July 2009 (UTC)

Problem with frame syntax

The Altviewer does not seem to find images specified with "frame". For example User:Eubulides/Sandbox currently has three images, but the Altviewer misses Image:Wikipedesketch1.png, I expect because that image is specified with "frame". Eubulides (talk) 22:33, 21 July 2009 (UTC)

There was no magnifying icon, no extra newline, so the regular expression broke. I've added \n* to places. Also, the source code is available at tools:~dispenser/sources/. — Dispenser 23:31, 21 July 2009 (UTC) [Update: It seems I had forgotten to symlink to it - 27 July]

Problem with wide image

For Georgia Institute of Technology the Altviewer mishandles the display of File:Georgia Tech from condo building at Peachtree St and North Ave.JPG, which is given a width of 700px (!) in the article. The large width causes the Altviewer to generate a table entry where the alt text and caption of the right column are superimposed on the image, making the alt text hard to read. Eubulides (talk) 19:36, 27 July 2009 (UTC)

This seems to be a rendering bug. IE simply cut off the text. Firefox splits the text out without the background oddly. The "scroll fix" is a greater pain than the problem. I'd suggest using the workaround of just highlighting the text you want to read. I will keep a lookout for possible fixes. — Dispenser 22:04, 27 July 2009 (UTC)

Mumbai makes Altviewer generate "substring not found"

tools:~dispenser/cgi-bin/altviewer.py/Mumbai causes the Altviewer to generate a "type 'exceptions.ValueError'" message highlighting line 46 ('html = html[html.index("<!-- start content -->"):html.index("<!-- end content -->")]') with the diagnostic:

<type 'exceptions.ValueError'>: substring not found
      args = ('substring not found',)
      message = 'substring not found'

/home/dispenser/public_html/cgi-bin/logs/tmpf5nHUa.html contains the description of this error.

Eubulides (talk) 17:31, 9 August 2009 (UTC)

The problem does not occur now. While it was happening, I double-checked it, and it consistently failed. Mumbai has not been edited since I made my comment, so it appears the problem (whatever it is) is at least partly independent of the content of the article. Eubulides (talk) 21:25, 9 August 2009 (UTC)

This seems to have been a problem with the squid cache somehow not including those scraping tags. The problem when away with action=purge and I have since changed the fetching mechanism to use the parser cache instead of the squid cache which has the side effect of re-rendering the interface elements. — Dispenser 22:23, 12 August 2009 (UTC)

Altviewer and videos

For International Space Station the Altviewer does not notice the video File:January 2009 ISS tour.ogg and does not display its alt text. It'd be helpful to show that video (or at least a placeholder for it, if there are technical difficulties with that), along with its alt text and caption; that way a reviewer can easily see whether videos and other media files have alt text. Eubulides (talk) 16:23, 10 August 2009 (UTC)

I'd added another section with the for video, but I really doubt that we'll get people adding descriptive captioning on the videos. You probably want to update the descriptions for the section. I would also suggest adding some bullet points somewhere about instead of these long explanations. — Dispenser 22:23, 12 August 2009 (UTC)

Shark image problems

For Shark the Altviewer does not find the alt text for Image:Tiger shark teeth.jpg or for Image:Parts of a shark.svg or for Image:Tiburón.jpg. It marks their alt text as being blank, but it is not blank. Eubulides (talk) 06:17, 14 August 2009 (UTC)

Seems like the tool lags a bit. The Shark alt text was added here, didn't show up for you ~24 hours later, and now shows up (for me at least). Similarly, I added alt text to Icos here and the tool is not displaying the changes yet. Shubinator (talk) 19:26, 17 August 2009 (UTC)

I am also observing problems with the tool lagging. Perhaps it could use a "purge" button? Eubulides (talk) 15:19, 19 August 2009 (UTC)

Big problems with Electron

I just now visited the Altviewer page for the Electron article and it misfired in a big way: it listed only seven small icons, and it failed to report all the thumbnails in Electron. The only images it reported (all with what it said were non-thumbnails with missing alt text) are File:Wiktionary-logo-en.png, File:Wikibooks-logo.svg, File:Wikiquote-logo.svg, File:Wikisource-logo.svg, File:Wikiversity-logo.svg, File:Commons-logo.svg, and File:Wikinews-logo.svg. Most of these seven images do not appear in Electron, and should not have been reported. Of the two images that do appear, File:Commons-logo.svg does have alt text, but the tool missed it. The only image that was handled correctly was File:Wikisource-logo.svg. Eubulides (talk) 22:54, 21 August 2009 (UTC)

You were viewing the alt text for Electron/ (the non-existent page). I've added code to check to make the page has history. — Dispenser 16:10, 30 August 2009 (UTC)

Not updating Alt text

I added Alt text to an article during an FAC but its not showing, only showing the previous ALT text.Teeninvestor (talk) 15:51, 30 August 2009 (UTC)

• I am observing the same problem. For example, currently the Altviewer view of Economic history of China (pre-1911) seems to be based on (approximately) this version dated about a day ago. When people are working on an article, feedback based on a day-old version is next to useless.
• If there's no way to get the Altviewer to get the current text, could it at least be modified to prominently show the timestamp of the version that it is analyzing, and to say how old that timestamp is? At least then the user will be warned that the tool isn't working well.

Eubulides (talk) 17:13, 30 August 2009 (UTC)

I have tracked to down to the squid cache returning old dated pages if "accept-encoding: gzip" wasn't sent which may be related to bug 7098. Hopefully, this fixes all the cache problems, otherwise I'll be looking to how to make purge work. I has also taken your suggestion and added the last modified timestamp to the bottom. — Dispenser 20:37, 30 August 2009 (UTC)

Wow, that's an obscure bug! (How do you figure out stuff like that?) Thanks very much for working on this. I have been relying so much on the Altviewer; it's a very helpful too. Eubulides (talk) 22:02, 30 August 2009 (UTC)

I ask on IRC and User:Para had suggested gzip header (presumably from the horrible GeoHack issue). Testing then isolated the header as the problem. An unintended side effect is that the page actually loads ~0.6 seconds (30%) faster (you can test it out by adding &nogzip to the url). I might have to add gzip support to GeoHack for faster page generation. — Dispenser 22:43, 30 August 2009 (UTC)

Unfortunately I am again observing the problem. The Altviewer page for Starvin' Marvin (South Park) currently shows only blank alt text, even though alt text was added over 20 minutes ago. The Altviewer's page says "This page was last modified on 31 August 2009 at 00:09" which, if UTC, means that it's more than a day old. As a result, I'm finding it difficult to respond to review requests like this one. Eubulides (talk) 03:17, 1 September 2009 (UTC)

I've asked on another IRC channel, they said it was the squid proxy caching the header/URL combination. The suggested fix is to change something in every request (like a timestamp) to bypass the squids. So I've reverted some of the "fixes" and implemented the simple solution. Hopefully this will fix the problem. — Dispenser 21:58, 7 September 2009 (UTC)

Altviewer is missing lead infobox images

Maybe this is due to the MediaWiki upgrade? but anyway, the Altviewer is currently missing some images that it should report. For example, its report for Cato June fails to mention the lead (infobox) image File:20070130 Cato June at Super Bowl XLI press conference.jpg. Similarly, its report for Battle of the Alamo fails to mention the lead image File:Alamo texas.jpg. Eubulides (talk) 03:38, 19 September 2009 (UTC)

Its been fixed in the new interface; not much point in fixing the old since I'm ditching the code in less than a week. — Dispenser 02:04, 20 September 2009 (UTC)

I've just installed this tool & its really useful but I have found the same problem re images in infoboxes. On River Brue & River Axe, Somerset the image in the infobox shows a red border & doesn't show the alt text which is used with it It may be the specific infobox.— Rod ^talk 13:25, 28 November 2009 (UTC)

The infobox uses |image_alt= instead of |alt=, fixed both of them for you. — Dispenser 05:27, 14 December 2009 (UTC)

Thanks I will try to remember to use image_alt= in geoboxes. In a related note I had some problems with this tool making the red dot on maps appear in the wrong place (examples include Brompton Regis & Exton, Somerset where the dot appeared outside the county boundary. I raised this at Wikipedia:Help_desk#.3F_problem_with_map_.26_coord_rendering, on IRC & ultimately at Wikipedia:Village_pump_(technical)#Strange_map_display_problem_in_FF. Having been through a range of attempts to fix this I found out I needed to remove User:Dispenser/alttextexplorer.js from my monobook to resolve the problem.— Rod ^talk 13:16, 14 December 2009 (UTC)

I'm looking into solving this. PS @Dispenser, i fixed a few other small issues with your script as well (like it overwriting potential global variables) 14:32, 14 December 2009 (UTC)

The issue with those relative positioned elements should be fixed. I also made a change that makes sure that if alttext==filename the red warning border is shown together with the set alt text. —TheDJ (talk • contribs) 21:08, 14 December 2009 (UTC)

Thanks for trying but I'm afraid it still gives the same error.— Rod ^talk 22:05, 14 December 2009 (UTC)

Javascript is cached WP:BYPASS —TheDJ (talk • contribs) 10:56, 15 December 2009 (UTC)

Thanks - now seems to be working as I would expect.— Rod ^talk 11:29, 15 December 2009 (UTC)

Altviewer test on the picture tutorial

I tried the new Altviewer out on WP:PIC and it had some real problems:

• It displayed a table that was waaaay too wide, due to the use of {{panorama}}. The table almost blew my browser out of the water.
• For the narrow images it displays alt text in a very tall thin stack, which isn't that useful.

Eubulides (talk) 05:36, 20 September 2009 (UTC)

I've since included a min-width: for the text boxes and the div scrolls now (not the best solution...) if they get too wide. — Dispenser 02:49, 9 November 2009 (UTC)

False alarm about unlinked image

The Altviewer analysis of Marshalsea says that File:Marshalsea.jpg is an "(Unlinked image)" but that image looks fine to me: it is marked with |link=|alt= and so should be considered to be OK, as it's marked as a purely decorative image, just as File:BigBenAtDusk.jpg is on the same page. Eubulides (talk) 16:35, 23 September 2009 (UTC)

Toolserver sapr issue

Thanks again for implementing the toolserver page to run the peer review script. When I made a template here link to it, I noticed a slight issue. If I urlencode a title with a space in it, the space becomes a +. It would be more user friendly if your script changed this plus back to a space in the text box. Example: [1]. — Carl (CBM · talk) 12:33, 29 July 2009 (UTC)

I noticed you removed the url escaping. I don't know what convention your script is using, but the following link does not work correctly to link to the peer review for the page A.26:

tools:~dispenser/view/Peerreviewer#page:A.26

Other page titles that include a period followed by something that looks like a hex code will have similar problems. — Carl (CBM · talk) 13:51, 29 July 2009 (UTC)

Now I get a Javascript alert box when I try this. I don't understand why you don't simply accept a URL-encoded title, as in

toolserver.org/~dispenser/view/Peerreviewer?page=PAGENAME

Then you can simply unescape the PAGENAME and there will be no issues at all. The Mediawiki section encoding is not a good substitute for URL encoding. — Carl (CBM · talk) 15:47, 29 July 2009 (UTC)

Actually, I see that this works, except that the URL rewriting is broken; try

http://toolserver.org/~dispenser/view/Peerreviewer?page=Food+fight

but there are UTF-8 errors with

http://toolserver.org/~dispenser/view/Peerreviewer?page=Kurt+G%C3%B6del

— Carl (CBM · talk) 15:49, 29 July 2009 (UTC)

So you've stumbled across the test code ;-). Well, I needed to make changes to the backend so it wouldn't look up file names with the ? character. However, I think it a dead path since I can't use the tools: prefix. I honestly don't think that the limitation imposed with the anchor encoded is a big problem as the user can just paste in the correct title. If is, I might look into just double escaping that part. — Dispenser 23:51, 29 July 2009 (UTC)

Ampersands and Peerreviewer

Peerreviewer doesn't like ampersands in article titles... when I try to put in either of the following, it runs a PR on Ratchet, instead of Ratchet & Clank: Up Your Arsenal: [[Ratchet %26 Clank: Up Your Arsenal]], [[Ratchet & Clank: Up Your Arsenal]]. It doesn't return anything with [[Ratchet & Clank: Up Your Arsenal]] Otherwise, this seems like a great tool. It also says that references are needed on articles which are already well-referenced, such as Neverwinter Nights 2: Mysteries of Westgate (PR). Otherwise, I've seen no reproduceable bugs, and it seems like a great tool overall. Thanks! –Drilnoth (T • C • L) 14:36, 29 July 2009 (UTC)

 Fixed I wasn't escaping the parameter before sending to the server. — Dispenser 15:18, 29 July 2009 (UTC)

Awesome! Thank you. –Drilnoth (T • C • L) 15:46, 29 July 2009 (UTC)

Image file links

Thanks for disambiguating some links, but your edit changed image links from File:Herminium monorchis (plant).jpg which is what it says nowadays on the image file page, to Image:Herminium monorchis (plant).jpg which I thought had been superseded. Is the change correct? Thanks, dave souza, talk 22:07, 30 July 2009 (UTC)

Again, see #Changing File: to Image: above. The commit notes specifically that File: does not superseded Image:. — Dispenser 10:29, 31 July 2009 (UTC)

Thanks, I've tried reading the commit notes a couple of times and end up puzzled. Changing File: to Image: when copying image names to an article seems a superfluous step, but there do seem to be some arguments for making that change. As this may have a bearing on the use of sreen readers by the blind I'll ask there if the issue should be made more explicit in guidelines. Thanks for resolving this particular question, dave souza, talk 09:40, 10 August 2009 (UTC)

Scroll box

Not really sure I followed your comments re: Template:Scroll box as used on the Template:Complete timeline of Macintosh models. If the height needs to be greater than 760 to work, it would seem the solution should be to increase the height to 761. Nothing in the Scroll box template applies to obscured image while scrolling. The userbility problem I have this way, is my browser window cannot open wider to display the entire table and text at the same time, requiring I shift the entire article to the right, losing the left navigation pane, which seems completely out of character for Wiki. I have never seen another article that requires scrolling off screen to see it. This article in particular has been this way since 2006. There must be another solution to fix whatever your problem is, which I am unable see. --Mac128 (talk) 01:11, 31 July 2009 (UTC)

I don't know what your screen resolution is but hopefully this simulation of the problem will help you understand. The usability problem is that I cannot see the upper third of the time line if I which to scroll horizontally. — Dispenser 10:24, 31 July 2009 (UTC)

Dablinks.py returning MySQL exception

I just now visited tools:~dispenser/cgi-bin/dablinks.py/Stanisław Koniecpolski and got a result back saying "<class '_mysql_exceptions.OperationalError'>" and bunch of other stuff. Eubulides (talk) 00:19, 3 August 2009 (UTC)

Same here. TomStar81 (Talk) 17:53, 3 August 2009 (UTC)

It will take about 90 hours (projected) until the reimporting of the English Wikipedia database is done. So if all goes well the tool will be back next week. Unfortunately, we don't have a global notice system to convey this information. — Dispenser 19:23, 4 August 2009 (UTC)

re: Your message

Hi Dispenser, Thanks for your message. In your edit on 6 August, you said ‘We have automated tools for this, don't do it by hand′.

Could you please explain how I can put this in motion. The Wikipedia articles I wish to check with Checklinks are in User:Marek69/sandbox11. I also have the same list in User:Marek69/sandbox10, which I have prepared to check with Reflinks.

Please could you advise me as to how I should proceed with this.

Thank you (in advance) for your assistance, which will be much appreciated :-)

Kind Regards 02:45, Marek.69 ^talk 02:46, 9 August 2009 (UTC)

It should all be explained in User:Dispenser/Checklinks/config. Since it looks like you've just dump your watchlist, you probably should opt for the monthly scans as the data takes an hour or two to generate. Since it unlikely that anyone is interested in that list. There isn't anything like that for reflinks, unless you count DumZiBoT. — Dispenser 08:34, 9 August 2009 (UTC)

Automatic Peer Reviewer

Just to inform you that your version of the peerreviewer script is being use by tools:~dispenser/view/Peer reviewer (see the about section). I hope this is alright with you. Also, somebody had ask to fix image: being change to file:. — Dispenser 16:10, 12 August 2009 (UTC)

Okay I've made the script check for File: as well. When I use the script, for some reason it returns different results every time I hit the "Load article" button. At the very least, I've noticed that this happens on the Samus Aran article. Gary King (talk) 02:34, 14 August 2009 (UTC)

As far as I can tell with the regular expression documentation it is a Firefox/Chrome bug. I've simplified down to a test case javascript:while(confirm(/x/g.test('xyx'))). It work as excepted if the g flag is dropped, which suggests that the pointer moves and resets at the end. — Dispenser 01:23, 16 August 2009 (UTC)

https://bugzilla.mozilla.org/show_bug.cgi?id=98409 — Dispenser 01:55, 16 August 2009 (UTC)

Graph article length over time

I'm wondering if you know of a tool to show the growth of an article over its lifetime. Like some sort of carting tool like Readablility, only it shows it over time.-- Patrick {_oѺ^∞} 00:42, 18 August 2009 (UTC)

The only tool I'm aware of is history flow. I doubt that we can run this on the toolserver due to the NFS slowing down read operations. — Dispenser 16:10, 30 August 2009 (UTC)

Dab tool not working for FAC

The dab tool at the top of Wikipedia:Featured article candidates is not working. Can you take a look? Thanks, Dabomb87 (talk) 14:29, 22 August 2009 (UTC)

Did you ever get a chance to look at this? Thanks, Dabomb87 (talk) 20:06, 29 August 2009 (UTC)

Initially I though you were referring to the s1 database outage that we were having. So the rewrite.script broke during the Solaris conversion/ZWS Upgrade which coverts %3F to ? has been worked around. Apparently the new regular expressions interpreter cannot handle (A|B|) and \w. — Dispenser 16:10, 30 August 2009 (UTC)

Thanks a lot. Dabomb87 (talk) 03:31, 1 September 2009 (UTC)

You're invited...

{{Wikipedia:Meetup/NYC/Invite/May 2009}}
This has been an automated delivery by BrownBot (talk) 02:03, 1 September 2009 (UTC)

backlinkscount.py

Hi Dispenser, you said that the source code was available for tools:~dispenser/cgi-bin/backlinkscount.py, where can I find it? I'd like to be able to tweak it in different ways, like right now tools:~dispenser/cgi-bin/backlinkscount.py?title=Template:Otheruses4 only returns 442. I'm guessing that it's not counting transclusion, does it count redirects? Thanks, — Sligocki (talk) 01:08, 8 September 2009 (UTC)

James Bond film series link changes

One of the new links doesn't work.--TonyTheTiger (t/c/bio/WP:CHICAGO/WP:LOTM) 03:27, 8 September 2009 (UTC)

[followup response] Thank for notifying me. As my followup edit says I have fixed the link on the page and configured the server to correct the other old ones. — Dispenser 02:49, 9 November 2009 (UTC)

Reflinks - tables, cite book, THANKS! :)

First, thank you very very much for this great tool!

I am having a bit of difficulty, though, with a couple of features:

• It seems to be (sometimes) changing Cite book templates incorrectly, trying to change "pages=" (total number of pages in book) to "page=" (a page number).
• It sometimes tries to help with table formatting, without complete success.

I see both errors when trying to use the tool on Gerald Ford‎‎. There are a few sources I would like to fill.

Thank you again for this tool! Please feel free to answer here, I am watching your talk page in hopes of an answer.- sinneed (talk) 16:22, 8 September 2009 (UTC)

I looked up documentation, page= referrers to an individual page, while pages= referrers to range of pages. The HTML fixes are to convert deprecated (and in HTML 5 removed) attributes. Some of the attributes are cellpadding, cellspacing, bgcolor, align, and they are converted as best as I can, but I'll look into work around your float:right/align=left mistake. — Dispenser 22:09, 8 September 2009 (UTC)

• Ah, thank you. I had obviously misread about the pages= option. Thank you! :)- sinneed (talk) 22:13, 8 September 2009 (UTC)

Coordinates log

Hey Dispenser, again one question concerning your log (tools:~dispenser/view/File_viewer#log:coord-dewiki.log). Is there a possibility to suppress all pages beginning with Vorlage:Positionskarte? These are only instructions how to use templates. --тнояsтеn ⇔ 16:12, 9 September 2009 (UTC)

Thanks for hiding these pages. One more question: can you please tell me what's wrong with the coordinates in de:Übelsberg, de:Artà and de:Palma? Thanks. --тнояsтеn ⇔ 12:06, 11 September 2009 (UTC)

People copy and pasted the GeoHack URLs on w:de:Benutzer Diskussion:Visi-on/Archiv09 and w:de:Wikipedia:Fragen zur Wikipedia/Archiv/2009/Woche 35. — Dispenser 05:25, 7 November 2009 (UTC)

Update

• Is it correct that you list dim and scale as errors in your log (tools:~dispenser/view/File_viewer#log:coord-dewiki.log)? Because it is larger than 4 MB now.

• Do you plan to update your other logs (tools:~dispenser/view/File_viewer#log:regioncheck-dewiki.log and tools:~dispenser/view/File_viewer#log:iwcoord-dewiki-CoordinateNO.log) on a regular basis?

--тнояsтеn ⇔ 14:19, 6 October 2009 (UTC)

• I added dim: support to GeoHack back in June, announced it on enwiki a month later, and sometime September it was added to the documentation. I added the check after I noticed that people were override the default dim: scale.
• I was unaware that these were being used, I was generating for tests data. The region check has issues.... And the interwiki coordinates was request by User:Docu for a bot (current specifications), I could run it weekly to if the data is going to be used.

— Dispenser 15:44, 6 October 2009 (UTC)

• I see. It is correct like this but working with the logfile is almost impossible now. I will ask the German geocoding team for a bot to remove scale parameters. --тнояsтеn ⇔ 19:58, 6 October 2009 (UTC)
• I just found them today by chance and think that they are really useful. If you could keep those logfiles updated every one or two weeks it would help a lot.

Thanks for you work! --тнояsтеn ⇔ 19:58, 6 October 2009 (UTC)

Once more

• Just to let you know: there has been an error in coord-dewiki.log since several days.
• If you are updating regioncheck-dewiki.log, please note that there is some problem with coordinates in France (they are marked as outside of Turkey).

Thanks --тнояsтеn ⇔ 20:26, 14 October 2009 (UTC)

• The database for S2 was down for some days
• The program has been rewritten since July and I do not intend to run it for anything other than testing purposes.

— Dispenser 05:25, 7 November 2009 (UTC)

coord-dewiki.log shows errors if dim contains decimal numbers which are used for ships and buildings in the German Wikipedia. --тнояsтеn ⇔ 14:20, 6 November 2009 (UTC)

I modified the code before so it would only warn if scale: value was not ten times of dim: value. The factor 10 is how GeoHack converts the values dim to scale. The code _dim:{{{dim}}}_scale:{{{dim}}}0 in de:Vorlage:CoordinateLINK does not handle decimals correctly. And scale: is now redundant to dim: I would recommend removing it. — Dispenser 05:25, 7 November 2009 (UTC)

And again

You asked for and i thought it's a good idea. But is the pagename really necessery? with the API comes this information implicitly. So finally, i d'like to remove this parameter. Visi-on (talk) 21:16, 25 November 2009 (UTC)

You'll need to be more specific next time. I’ll assume that you ment when User:Docu asked that pagename= be added to at w:de:Vorlage Diskussion:KoordinateURL (now recreated at w:de:Vorlage:CoordinateLINK) for my db log. Its an optional parameter as far as I'm concerned. Tools that rely on pagename= all should have fallbacks, so I'll try and explain them.

GeoHack will make an attempt to get the pagename from the HTTP referer, but it only works if it match http://\w+\.wikipedia\.org/wiki/([^#?]+) expression (no /w/index.php or secure server). If GeoTemplate is setup correctly then links that require {pagename} (e.g. "All coordinates on pagename") will be hidden by CSS.

Now my database builder (coord-*.log generator) doesn't use the MediaWiki API (far too slow), but uses native SQL. It does not JOIN the extenallinks and page tables together, since in certain conditions (enwiki) it was a performance hit. This precludes the use of page namespace and page title from being used in the script. This isn't an issue since they're only useful for the logs and the fallback ?curid=### is good enough. — Dispenser 05:45, 26 November 2009 (UTC)

GeoTemplate/venus

Hi,

I created {{GeoTemplate/venus}}, based on your work on {{GeoTemplate/mars}}. Could you check it, please? It's called by pages using {{Venus coords and quad cat}}. Cheers, Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 15:38, 21 September 2009 (UTC)

I've expanded the documentation of the Template:GeoTemplate/sandbox#Location map and have implemented the multiplication technique on Template:GeoTemplate/moon and added mention of the Equirectangular projection requirement for maps images. Typically I test the corners to make sure things align right. I don't have much time to look to see if your implementation is correct. Sorry. — Dispenser 03:29, 22 September 2009 (UTC)

You don't seem to be a big fan of RTFM or you would have realized that WP:DOC doesn't play nice with GeoHack, or that changing the icon require typically changes the offset requiring recalculating the div offset layer. :-P Also, if you have any design idea, feel free to add them to the collection growing in the sandbox. — Dispenser 03:48, 23 September 2009 (UTC)

I miss PDFbot

Hey Dispenser, I noticed PDFbot hasn't run since June. Are you planning to run it again? I liked what it did, and I have an article that could really use it. Jayjg ^(talk) 02:51, 29 September 2009 (UTC)

Don't mean to bug you, but is there any chance it is coming back? Jayjg ^(talk) 01:05, 13 October 2009 (UTC)

I'm pretty busy with other project, but I intend to recover the files from the server and move it over to the Toolserver. If you really need to use you can try using the web port. — Dispenser 21:41, 16 October 2009 (UTC)

Thanks for getting it working again. Jayjg ^(talk) 01:50, 7 December 2009 (UTC)

Wikis Take Manhattan

{{Wikipedia:Meetup/NYC/Invite/WTM3}}
This has been an automated delivery by BrownBot (talk) 20:54, 4 October 2009 (UTC)

Lyon page layout issues

Hi Dispenser- I saw your name in the history for Help:Table, looked at your user page, and thought I'd throw you an FYI regarding my somewhat tragic attempts to fix the page formatting code in the Lyon article. Do you think I'm on the right track with my observations on that talk page? Thanks in advance for even a glance at the problem. Eric ^talk 17:12, 7 October 2009 (UTC)

WP:BUNCH. You really should remove most of the images and maybe setup a page on commons with the best of them. I think there was some page on that, found it WP:IG. — Dispenser 02:49, 9 November 2009 (UTC)

Yeah, I know the bunching template, but didn't have much success with it on that article. I agree on the gallery idea, but haven't yet taken that much ownership on the article. Thanks. Eric ^talk 03:58, 20 November 2009 (UTC)

Tools review

Hi Dispenser, I quickly whizzed through the tools as you asked and tried them on an article that I know has minor problems. Excuse the bullets!

• Altviewer - Great
• Checklinks - Great, a little slow to load but I guess that it is doing a lot, is the second checklink tool listed the same one?
• Dablinks - Great
• Peer reviewer - Very good, sometimes throws up strange results (nbsps needed between non-units). Seen a lot in aircraft articles where numbers and letters are used together, Boeing 737 NG for instance, not a problem.
• Readability - Not seen that before, a bit technical for me but looks very useful.
• Reflinks - Complicated for me at first glance, maybe needs an expanded instruction guide. I am wary of bots or firing one up due to lack of knowledge!
• Pywikipedia and Geodata - Way beyond me!

Hope that helps, I have added a link to the tool page to my user page, they will all wonder how I am spotting errors!! Cheers. Nimbus (Cumulus nimbus floats by) 00:48, 9 October 2009 (UTC)

The PDFlink template

Hiya, Dispenser. I'm just curious about something you mentioned on the PDFlink talk page. In your support for the delinking of the "PDF" in the template, you said that you had "actually been delinking some of the hard links with PDFbot". Can changes like this be made to a bot without getting a third approval? (Of course, it's a fairly moot point now that the template's been changed.) I was going to ask someone over at the BAG, but then I thought 'no'... I'll check with Dispenser first. That's some pretty great work you do, and believe me, I am fully clueless when it comes to the programming level. I sincerely revere such work.
 —  ^{.`</sup>^) Paine Ellsworth<sup>diss`cuss} (^^`.  20:23, 14 October 2009 (UTC)

It more or less a loophole in the third BRfA to use the commonfixes library, but I still listed at one of the items it would do. I also use this library with my tools. — Dispenser 19:06, 7 November 2009 (UTC)

coord dim: parameter

In your edit summary you say that the on-screen diameter is exactly 10 cm at 72 dpi. However, most mapping services only have a finite number of scales they can display, so there must be substantial round-off at some point in the process. In most cases, therefore, the size of the onscreen viewing circle does not exactly match your calculation. Can I add "roughly" back into the documentation? --Stepheng3 (talk) 10:12, 15 October 2009 (UTC)

I wasn't thinking about the map services when reading the text. Feel free to revert that edit. — Dispenser 21:41, 16 October 2009 (UTC)

Suffixes for dim:

I'd greatly appreciate it if you could implement the 'km' suffix for dim:, and also arrange to be able to tolerate an optional 'm' suffix: this would make dim: much more self-documenting: 'dim:100m' in the context of a large building, for example, is pretty much self-explanatory. I can't see any need for any other SI suffixes at the moment: as said on the other talk page, 61cm is perfectly well codable as 0.61m, and things like 3000km are just fine for all earthly applications, in my opinion. -- The Anome (talk) 11:17, 16 October 2009 (UTC)

I've added a hack that will replace 'km' with '000', this needs to be cleaned up since it doesn't handle decimals well. — Dispenser 19:06, 7 November 2009 (UTC)

Speedy deletion tip

When you want to request speedy deletion of a redirect page, please put the speedy deletion template at the beginning of the page, before the #REDIRECT directive; otherwise it is difficult for the reviewing administrator to find and review the request. Thanks. --R'n'B (call me Russ) 13:06, 19 October 2009 (UTC)

I thought that part of the text would be preserved as part of the summary. Undoubtedly, deletion needs to be handled better. — Dispenser 19:06, 7 November 2009 (UTC)

An AD Campaign

Hey there Dispenser,

Im going on a bit of an AD campaign for your tool : CHECKLINKS. I introduced a template ({{Dead link header}}) that was aproved (here) and asked someone to help me with the bot that will add this to articles. Just FYI. --Tim1357 (talk) 21:56, 20 October 2009 (UTC)

I think the tool needs less of an ad campaign a much better interface. It is possible to do many things, but it is currently too time consuming or requires too many clicks. — Dispenser 19:06, 7 November 2009 (UTC)

Well I remember you saying that the page gets something like 80 hits a day. (I re-preposed it here) Tim1357 (talk) 02:45, 18 November 2009 (UTC)

{{Dead link header}}

Hey Dispenser,

I created a new template: {{Dead link header}}, that provides a link to your tool , Checklinks. I would greatly appreciate it if the tool removed all instances of the template when it is run on an article. Thanks, Tim1357--(what?...ohhh) 13:43, 25 October 2009 (UTC)

It will now remove the template if there are no {{dead link}} tags in the article. — Dispenser 19:06, 7 November 2009 (UTC)

Thank you

Looks good from here; thank you so much for all you do. SandyGeorgia (Talk) 15:20, 27 October 2009 (UTC)

Reflinks at Brazil

I swear I feel as if I have lost an arm when I can't Reflinks an article. Brazil kills Reflinks when I run it. Just an FYI. Thanks again.- Sinneed 19:42, 27 October 2009 (UTC)

I've determined it's a bug with OpenSSL and python on the Solaris servers. I've added a hack to skip all https: url and opened a ticket, TS-385. — Dispenser 02:49, 9 November 2009 (UTC)

Re: Peer reviewer maintainer

There has very little activity with User:Gary King/peer reviewer.js. Would you mind if I added a notice asking for a new maintainer of the peer reviewer script? — Dispenser 05:48, 2 November 2009 (UTC)

Okay, I don't plan to do anything with it anyway. Gary King (talk) 06:46, 2 November 2009 (UTC)

You're invited!

{{Wikipedia:Meetup/NYC/Invite/November 2009}}
This has been an automated delivery by BrownBot (talk) 03:20, 3 November 2009 (UTC)

Thanks

Hi. Just a small note to thank you for providing the http://toolserver.org/~dispenser/logs/coord-nlwiki.log page. I discovered it yesterday so I could finally correct all kind of errors very effectively. Great job. Michiel1972 (talk) 20:20, 12 November 2009 (UTC)

Lots of people seem to like the tools:~dispenser/view/File_viewer#log:coord-nlwiki.log interface, BTW. — Dispenser 06:12, 15 November 2009 (UTC)

Google Mars issues

I've been trying to get the Google Mars link on Template:GeoTemplate/mars to work with the Dawes Crater testcase on Template talk:GeoTemplate, which has been hanging. I've found that longitudes less than -180 cause Google Mars to display a blank map. (Dawes Crater is at 322 degrees west longitude on Mars, which is why it triggers this issue.) Could you stick a modulo-360 function into the calculation of {londegdec} such that the range -360 to -180 would get mapped to 0 to 180? --Stepheng3 (talk) 02:30, 19 November 2009 (UTC)

PHP's modulo function truncates the decimals. It would have to worked around with something like while(lat < -180) lat+=360;. I would like to research this better to understand if it should also be converting them for values greater than 180. I should be able look into it after we've completed the move off the stable servers. — Dispenser 23:23, 19 November 2009 (UTC)

I tested values >180 and they seemed to work. I think a workaround of some sort would be worth adding immediately, even if there's a move immanent. --Stepheng3 (talk) 23:43, 19 November 2009 (UTC)

By the way, I'm wondering how well {zoom} works with Google Mars. Does Google Mars really use the same zoom-to-scale function as Google Maps despite the fact that a degree of arc is much smaller on Mars than it is on Earth? --Stepheng3 (talk) 23:54, 19 November 2009 (UTC)

Border?

Is there a reason PDFbot adds the border width, such as here? --NE2 21:54, 20 November 2009 (UTC)

Old version of the commonfixes library. I will update it when I get around to running more than the template error cleanup. It's harmless since the wikitable CSS class overrides it. — Dispenser 22:05, 20 November 2009 (UTC)

Spellchecking is online

Hi most indispensable Dispenser. We have now added the spellchecking fix to MediaWiki:Common.js/edit.js. So it's time to bypass your browser cache and you will have spellchecking in the edit summary field.

--David Göthberg (talk) 17:11, 25 November 2009 (UTC)

featured article tools at DE

Hi Dispenser, I just answered your statement regarding featured article tools over at de:Wikipedia_Diskussion:Kandidaturen_von_Artikeln,_Listen_und_Portalen#English_.22Featured_article.22_tools. Regards, --X-Weinzar (talk) 22:26, 30 November 2009 (UTC)

something for checklinks

hey there dispenser, I just wrote some script that takes links, and dates fed into it, and returns the internet archive url that is closest to that date (within the range of 4 months from the original date) so it would work like this

getArchiveURL('http://Iamdead.com', '20080304')

that wold return the archive closest to the date 03/04/2008. It only works if the date is in the format yyyymmdd. I was thinking you may be able to use it in checklinks, to take some of the repetitive work away from the user. Here is the full code, if you want it.

Code

 import urllib2

import datetime
def dateranger(numbers): #this module returns the range of dates to search for. The internet archive api takes this, and returns the url closest to the center.
    daterangehigh = datetime.timedelta(weeks=8)
    daterangelow = datetime.timedelta(weeks=-4)
    year = int(str(numbers) [:4])
    month = int(str(numbers) [4:6])
    day = int(str(numbers) [6:10])
    accessdate = datetime.date(year,month,day)
    accessdatehigh = str(accessdate + daterangehigh)
    accessdatelow = str(accessdate + daterangelow)
    return accessdatelow.replace('-','')  + '-' + accessdatehigh.replace('-','') 
 
def getArchiveURL(archiveURL, date): #the following was borrowed from weblinkchecker
        archiveURL = 'http://web.archive.org/web/%s/%s' % (dateranger(int(date)), archiveURL)
        try:
            f = urllib2.urlopen(archiveURL)
        except urllib2.HTTPError:
            # The Internet Archive yields a 403 error when the site was not
            # archived due to robots.txt restrictions.
            return None
        except UnicodeEncodeError:
            return None
        data = f.read()
        if f.headers.get('content-encoding', None) == 'gzip':
            # Since 2008, the Internet Archive returns pages in GZIPed
            # compression format. Unfortunatelly urllib2 doesn't handle
            # the decompression for us, so we have to do it ourselves.
            import gzip, StringIO
            data = gzip.GzipFile(fileobj=StringIO.StringIO(data)).read()
        if "// FILE ARCHIVED ON" in data:
         y = re.search('(?<=//\sFILE\sARCHIVED\sON\s)(\d+)',x)
         return 'web.archive.org/web/' + y.group() + archiveURL
        else:
            return None

I hacked something similar to this during Thanksgiving Weekend using JavaScript and python. As I have mention before, I would like greater automation in the tool and easy of use. Unfortunately, with classes right now I do not have the time to code, hopefully it will change in a few weeks. Anyway, in the grand scheme I believe the problem is that webmaster lack information and awareness to prevent linkrot. I will add a utility for webmasters (a la Google Webmaster Tools) allowing viewing of Checklinks' data by domain rather than article. Then somebody needs to create a How To, similar to the revamped WP:Linkrot, explain how to use 301s, 302s, 303, 307s, and 410s, discussing different URL scheme including short URLs, and ways of avoiding linkrot for webmasters to read. And template/sample emails need to created so wikipedians can easily pester webmasters. — Dispenser 04:58, 14 December 2009 (UTC)

bot request

I'd appreciate it if you would respond to comments I made at Wikipedia:Bot requests. Cheers, --Stepheng3 (talk) 04:31, 4 December 2009 (UTC)

Too late--the discussion has already been archived. I find it disheartening that you quashed my request the way you did. --Stepheng3 (talk) 21:45, 13 December 2009 (UTC)

Secure server

Hi Dispenser. Thanks for the code you showed over at MediaWiki talk:Sharedupload-desc-here#Link to secure server. It inspired me to make the template {{sec link auto}} and write Wikipedia:Secure server. I am now applying {{sec link auto}} all over the place, including the Main Page. See Wikipedia talk:Secure server#Sister project links. And we have even changed how the secure server log in looks. :))

--David Göthberg (talk) 03:17, 24 December 2009 (UTC)

GeoHack

Is it safe to change the url from stable? Noticed ~geohack is working, just dont know if it's ready or not. Q ^{T C} 00:36, 31 December 2009 (UTC)

No, I am planning on doing some work which would cause breakage. When ready, old GeoHack URLs will be 301 redirected per TS-368. — Dispenser 16:57, 31 December 2009 (UTC)