I decided that it would be useful (to me) to record some of the thoughts and ideas I have about the Wiki world, and certain actions I take, as they occur. In line with the open nature of the project I'm having a go at doing it on-wiki, I'll see how it goes.
This is an experiment in more ways than one:
Firstly, it's my début in blogging.
Secondly I will be doing it using mirror threads, which is itself something I just started working on.
Thirdly it is slightly more "meta" than I usually go on-wiki
Fourthly it may be of interest to no-one but me (which is fine), but if so I'll may switch to a local document.
Working on the templates for Mirror threads - most of it is put-off-able until the bot is running, but {{Mirror}} itself is centring text and I need to probably move from or mod {{navbox}}.
A little User talk page traffic.
I made a progress box (see right) for all undated categories a few days ago, unfortunately it was unable to be accurate without carnal knowledge of subcategory counts - essentially it corresponds to User:SmackBot's backlog - I created Category:Counter categories to deal with this - there will still be a few dozen undated sub-categories, but the headline figure will no longer be out by 2051. Some of the counter cats still need to be created - SilkTorque's user page was catting as one, which was amusing. There are also some anomalies around the "as of" cats, as they work on a year/month hierarchy. All the dated cats needed a null edit.
Ran the usual cleanup run for SB, and its Runescape Wikia run was just kicked off. Using a patched AWB for WP to try and get timings for search and replace as it is stalling on long pages. After 2G of log files, I'm little the wiser, seems like a .Net/windows scheduling problem.
Also ran SB against other namespaces - largely manually, talk for the technical, and user for {{User draft}}, picked up one page in File: space, which is oddly few, and removed a demospace parameter that was fouling categorisation. Nice to see the WikiProject names lining up in the banners, all except 5 projects out of hundreds have reasonably standard banner name, and four of those five have cromulent redirects.
* created a short perl script to make special settings for talk pages - played with different perl styles that I rarely use and came up with a nice 5 liner that runs before you can blink.
Added the "Technical" and "page number required" clean-ups to the dated structure, I have been dating these templates for years so they should slot in, although the tech one is mainly talk pages so there will be a September 2010 bulge - {{DMCAT}} created for templates that go on article and article talk, as was always intended from day one.
Added a handful of dated templates to SB and also Infobox boxer - need a more scalable way of dealing with that aspect of clean-up - I'm thinking about an AWB component that caches template names and redirects - will need "R with possibilities" compatibility.
Mirror needs to use a 1= for it's first parameter, in case of "=" in the text, this is a shame as it makes usage slightly less intuitive. Of course Mirror Bot should fix it up, but I think the instructions will need to mention it.
Found a page in the "Invalid ISBNs" category! that's a blast form the past. Also a bunch of old-syntax portal pages have re-emerged.
Created the dated "technical" cats, semi-auto. Still need to do the detail work on counter cats.
Would really like to get rid of the remaining "All pages with problem X" cats, they clutter the hidden cats on an article.
Fixed up Multiple issues to cat quote-farm - but no "all" category, thank you.
{{No definition}} added to SBs list of 552 dated templates. Only used in about 4 or 5 articles. Backlog is about 4-500 but many of these are category lag and are actually fine.
Fixed a SB bug relating to the transition from article issues to multiple issues - this was invisible before due to the category lag backlog. Should allow SB to fix a good percentage of the 196 outstanding items.
Manually dated tag in Aorist - this protected page has resisted automated attempts for a while.
Developed {{Monthly clean up category}} so that empty cats over a month old nominate themselves for deletion. (Last month's category might be useful to copy to this month's.)
Added {{ASF}} to SmackBot build 521 I think. The rump is down to 160
Reflist run, and Wikia run.
Not sure what to do about clean-up templates like {{Notability|notability=September 2010}} this type of error is rare but part of the "using non-existent parameters" class of errors that causes a lot of problems.
Implemented yet another mis-spelling of "date"
Decided to use an "advanced" rule to remove clean-up categories that are placed explicitly - it will skip if Ambox, Fix or Multiple issues/messages are present to avoid removing cats from foolishly substed templates.
* That seems to work nicely.
Would be good to tweak the rules for template redirect replacement not to be broken by html comments.
{{Dating maintenance categories progress box}} looks great now that it is reflecting almost accurate figures. Took a little work, the counter cats to be created (71 of them) and as a side effect sent me off to create {{expand Uzbek}}.
Mainly concentrated on getting the rump SB articles down to zero! The usual template tweaking, added a couple more expand language templates, created {{Smack me}} that puts articles in a clean-up category, and gets removes by SB's dating/GFs run.
Helped Dr Blofeld with the Chinese county articles.
Aorist popped up again - AWb seems to have problems with it.
Dated the cat Wikipedia pages needing page number citations
Already articles are popping up creating a new SB rump - subst:'d merge templates of course and "As of" with a badly formatted parameter (As of gets special treatment so there aren't as many cases dealt with). Newer errors that can't be systematically dealt with can at least have educational messages left for the "perps".
Developed {{In Our Time}} which provides a link and a "listen again" link to the 330 archived programs on the BBC. Left notes on about 100 talk pages.
Investigated the use of {{Bots}} for the first time for months. There are over 2000 pages transcluding the template, about half in user-space. This is far too many, the template is intended as a stop gap (except for user talk pages). Part of the problem relates to the use of the template on navboxes which are transcluded, where the template is mainly targeted against DashBot, and on citation templates (doi) created by Citation Bot, where it is the target. Also targeted are interwiki-bots and on the red-Links project for no good reason I can see AWB.
Did a lot of work trying to analyse interwiki language links across Wikipedias, with some success. Still a couple of bugs on picking up foreign language dates and a rare bug picking up incorrect transaction numbers.
Ah.. Noddy (character), the little man with the red and yellow car. Noddy (character) ... How come we don't have a decent coverage of this literary great? Still I made some links and some empty sections....
The coverage of the Queens Awards is woeful, I managed to fill in the awardees for the last five years, and am hopeful that I can get the rest from the London Gazette. I have the 2005 individual award list courtesy of the QA office. Also need to look at the QA for volunteers.
Found some deprecated parameters from some infoboxes seem to have crept back, they can be removed fairly easily, which is worth doing proactively because people will waste time filling them in otherwise. More plans for cleaning infoboxen in the pipeline.
"Bot battles" are an irritation just now, but we should be able to get them resolved easily enough.
AN seems to be calming down - the message that all the complaints were addressed some time ago seems to be percolating. Thanks for the many emails of support.
Did a SmackBot run of mis-capitalised titles - an error had crept in with WSOP, easily cured and picked up about 30 existing errors in articles while cleaning up
Fixing up Navboxes with the wrong "name" parameter Template:Ligne_de_la_Côte_Fleurie woul dhave opened "Sheffeld stations" if using the vde..
Have more samples for high risk BLPs might run that later tonight
Reflist really needs to come off AWB and run standalone.
Created a patching program so that I can maintain two copies of AWB - one for SmackBot with the restrictions CBM wnts, and a real one for me. Just need to write the patches...
Made a start on DNB migration
Resumed blogging - even if it's invisible until I fix up the template. Maybe Lugnuts might help with that.
26 October Uh.. we didn't think this one though.[edit]
Editing restrictions.
Wikipedians, are by and large, lovely people. With the best intent in the world. However many of them don't fully understand Wikipedia - I certainly don't myself - nor do they need to to make fantastic contributions. And when I do, I cannot necessarily bring all the facts to bear at the right time.
So my "not-punishment" - for doing something which was certainly very terrible, but nobody was quite sure what is, seemed fairly mild, although I could see it might be more of a headache than was apparent. The restriction - not to make changes which don't affect the rendered page. Who would want to make such changes? Well I glanced at two editors history - the first two I picked up, the one who suggested the remedy, and the one who enacted it. Both had done so within a handful of edits, one in the first edit I looked at. Still I mentally shrugged, irritating, no doubt, but just a change in the way I work, surely. I could not be more wrong. Within a few edits I had come across a typo in an HTML comment. I am not allowed to correct this. I started to see just how onerous this is:
I am not allowed to change, add or remove HTML comments.
I am not allowed to align an infobox.
I am not allowed to remove unused parameters from templates.
I am not allowed to edit any page with leading blank lines (who is going to check first?)
I am not allowed to remove surplus space from the middle of text (six spaces between consecutive words for example).
I am not allowed to add, remove or change a default-sort.
I am not allowed to date a maintenance tag
I am not allowed to add, change or remove any parameters in a Persondata.
I am not allowed to correct deprecated HTML, or replace it with wikicode
I am not allowed to change invalid HTML unless it visibly breaks something
I am not allowed to remove redundant cruft to reduce pagesize as I did for the then huge Spanish Football 2009 page (I think it was), making it possible to load.
I am not allowed to make stuff easier to edit by use of templates like this
I am not allowed to tag stubs as unreferenced.
I am not allowed to fix an incorrect piped link.
I am not allowed to disambiguate.
I am not allowed to correct or update a url.
I am not allowed to correct or update an interwiki.
I am not allowed to revert myself if I do do one of the above.
That's just what has crossed my mind so far. Happy editing!
Life goes on, SmackBot's main task is running, got the October run through round about month end, November is running now
Built the CBM restricted version of the new AWb (although I am probably many releases behind) - that means I can run SB regularly which is how it should be, running daily shows problems quickly, means I can do the manual fixes and improve the algorithms.
Did a bunch of date errors in the last few days, got about 1800 // dates to do yet and a few thousand typos in the WP 0.8 articles
Want to get back to porting old DNB articles, it's fairly nearly done.
Also need to get the BLP risk reports automated. Again distraction at ANI and recovering the state of the code from Septemeber for other jobs has made everything difficult.
The protein stubs need categories there is a good resource at QMC we need to ask for it to be released CC-by.
Reflist task needs work and ideally moving from AWB to become stand-alone and a bit smarter. Of course that looses the GFs to an extent.
The migration of {{Monthly clean up category}}to{{Monthly clean-up category}} proved not as simple as expected. It was really a mistake to give in to pressure on that - it's not grammatically wrong and even if it was,despite being used 2-3,000 times, no human should ever have to read it -Femto Bot creates the cats, and they self-nom for deletion when they are empty.
Another potentially big project is creating bibliographies - User:John Carter has pointed out how useful these can be to new users.
The standardisation of banner names continues slowly, WP Film came on board, Biography looks promising, although US Roads turned up there to protest. That would leave just three "odd" banners, which is much better than the hundreds we had a year ago. Roads and Military history seem happy for the redirect to be used - as long as they don't need to do anything and their watch-lists don't get spammed, which is reasonable - they don't have that large numbers of pages less than 100k each. I'm expecting it will take 3-5 years to substantially standardise the banners.
Also tricky is an error fixing BRFA where I can't find any examples of the error (repeated clean-up tags) to fix for a trial.
Also just recently created W:WikiProject Stamford - partly to get some of the fascinating story of the town on-wiki and partly to find out how extensively we can create content - Stamford has many centuries of recorded history, and figures in the stories of the Danelaw, Magna Carta, Wars of the Roses and the Civil War to name some highlights.
Got Femto Bot tasks running of a central scheduler, still primitive but simple is good.
SmackBot's main task is moved to perl which enables it to keep very slightly better track of what it has changed, but throws the baby out with the bathwater in some ways
Femto Bot continues to do sterling work, relieving me (in particular) of various tasks, but there is still a problem writing large pages - the US project recent changes in particular. Need to test if transcluded sub-pages will work with "related changes"
Also a problem is intermittent "bad tokens" on the API.
For Mirror Bot I need to get toolserver access, I believe that it could prove very useful, and not be a good idea to run from here. I applied years ago but it all went vague and wibbley.
My O'Reilly's "Perl DBI" arrived today (well yesterday actually) which will help me to use a database from perl.
Cleaned up Category:Empty categories, mainly by repopulating the categories emptied in error, a useful exercise in many ways.
Glancing at my talk page there's a number of outstanding things to be cleared up, but nothing urgent, priority should go to the bots, then DNB, the maybe enzymes.
I also noticed that Wikimania publicity is starting. It would be good to go this year, especially if I had a lap-top to take with me.
