User:Pelagic/sandbox/j/2020/04

April 2020

Friday 1 May
Yeah, I know, I should start a new month.

Notifications
Good grief, I have 29 notifications on mobile and 10+29 on desktop?? Except the mobile one changes after I open it even though I haven't read or ticked off any of the items? I wonder whether less than 80% of those come from Structured Discussions threads on MW?

Names
The normal western approach of encoding names as ( given names, family names ) or even ( title, given name, middle names, family name ) is just terrible. It leads to situations where the payroll database with the subjects' legal / banking names is unmatchable to the general directory (e.g. AD) having their preferred display names for email and interpersonal use.

There was a defunct proposal for FOAF:


 * familiarName - the name used in familiar situations by friends, e.g. Ian
 * informalName - the name used when referred to in informal situations, e.g. Ian Davis
 * formalName - the name used when referred to in formal situations, e.g. Mr. Davis
 * fullName - the full name as expected to be used on an identifying document e.g. Mr Ian Robert Davis

I wouldn't put Mr in the fullName, and I'd use different labels, but this comes close to some common use cases. Consider a fictional Mister Tony Baloney (his nonna calls him Marco):
 * Hi, Tony, …
 * 1) Dear Mr Baloney, …
 * 2) Pay to: Mark Anthony Baloney
 * 3) Regards, Tony Baloney
 * 4) Sincerely, M. Anthony Baloney, M.Sc., FRS, OAM

For a first-name sort, should he be ordered under M, A, or T?

The different forms start to increase when you consider fullLegalName versus longNameWithInitials. And accountName for banking could be completely different again.

A person's legal name in a particular country or jurisdiction might be mangled to local conventions, and be different to their culturally-aware full name. If Gabriel García y Márquez had lived in Australia, his legal name could have been plain "Gabriel Jose Garcia" (you can have middle names, but only one surname, and no accents, thanks). Would his familiarName be Gabriel or Gabo? Maybe nickName for the latter.

When we get into honorifics, things get more complicated. Is it Nanak, Guru Nanak, Shri Guru Nanak-Ji Sahib? Or some other subset of the last?

Then there are historical persons who may have a toponym like Leonardo of Pisa. Or they may have a geographical qualifier a noble title, like one of my favourites, Bernard Lacepede, more properly Bernard-Germain-Étienne de La Ville-sur-Illon, comte de La Cépède.

And the type of properties you want for marking up display text may differ from what you'd use for, say generating a letter. "Dear Citoyen La Cépède"? "Dear M. le Comte"? ?

Object vs literal properties in data
OWL encourages vocabularies to avoid using a single named property in both 'ObjectProperty' and 'DataTypeProperty' styles. However earlier usage, notably in the Dublin Core community, does just this. Each FOAF property is either an Object Property or DataType Property. 

Open tabs

 * Phab
 * T248061 Remove IE 8 from basic support. Interesting info there about the approach to browser support.
 * Compatibility
 * T63801 Wrap MediaWiki footer to HTML5 tag (2014–2015) declined, ARIA attributes are more widely supported.
 * T25932 Allow use of semantic HTML5 elements in wikitext. Has a handy list of HTML elements that can and can't be used.
 * Wikipedia Signpost/2020-03-29/Discussion report. I never got around to commenting on WP:VPP, though I did on WP:VPP.

Essay
Pete Forsyth, "Trusting everybody to work together", The Signpost 2020-04-26. I've pulled out a few quotes that I particularly like.

This element—facilitating reader insight into writers' actions and motivations—may at the time have seemed an ancillary benefit, secondary to the need to support a burgeoning community of writers and editors. But transparency to readers has become a key component of Wikipedia's identity, and as Wikipedia's star has risen, many of us have come to expect greater transparency from more traditional publications as well.

Diffs, history, and other "advanced" features aren't just for editors and involved community's members. They also benefit readers. — In addition, anyone consulting the encyclopedia also has access to this information. If a sophisticated reader familiar with the features outlined above sees something in an article that doesn’t ring true to them, he can learn something about how the text evolved to that point, including what the article's authors argued about, or what sources they considered but dismissed.

Veteran Wikipedia author Jim Heaphy published an essay outlining a number of reasons why he uses the desktop interface while editing with his smartphone, instead of the specialized interfaces created by the Wikimedia Foundation, later that year. 

In a traditional context, readers looking for data to inform trust were typically limited to familiarity with a publisher's or a writer's reputation. Did you read something in the New York Times? You could probably trust that was true. Did you read it scribbled on a bathroom wall? Not so much. Wikipedia, as a product subject to continuous editing by almost anybody, takes a different path; it does not aspire to the kind of reputational trust enjoyed by the Times, but the eight software features in this essay separate it from the wholly unaccountable bathroom wall.

A reader literate in the scientific method is better equipped to evaluate a scientific study than one who has to rely on external authorities; a television viewer well-versed in the techniques of video manipulation or rhetorical trickery will be less susceptible to deception.

To fully appreciate the value of Wikipedia, a reader needs to consult features like talk pages and edit histories. As Wikipedia has grown, and as MediaWiki and similar software has proliferated across numerous websites, an ability to work with these software features has become a core part of information literacy. They should be taught in our formal educational institutions, and curious readers should investigate them on their own.

Time functions
Time has good doco for reference. Ideally I'd like to build something with magic words or parser functions that doesn't depend on a specific wiki's collection of templates. Magic words affect caching, so I'll focus on PFs instead.

Parser functions: #time, #expr.

Putting local time in sig
For reference  without custom sig. produces

After customisation, this inserts as Pelagic (talk) – (14:13 Sun 26 AEST) 04:13, 26 April 2020 (UTC)

Sure, I could add logic for DST with a hard-coded start–end, but is it worthwhile compared to just editing the sig twice a year?

Saturday 25
Anzac Day



Matcha
According to this post, maccha powder is prepared into either usucha 'thin tea' or koicha 'thick tea'.

Sidebars
Sidebar items for the 15 wikis having over a million articles (according to the list at the foot of en-wp Main Page). The focus here is what's included, not the order, grouping, or exact name. (1) "Bar"

(2) Italian sportello informazione looks more like our helpdesk rather than reference desk. Not sure about the Chinese one. I grouped Vietnamese "Help desk" here also.

(3) "Forum"

(4) "Tavern"

(5) "Introduction", not sure how similar this is to "first steps" on other wikis. It's also called "introductory course" on-page, looks a little like en-wp's "Wikipedia adventure"?

(6) "Announcements"

(7) "General discussion"

Articles (millions): sv 3.7, pl 1.4, jp 1.2, ar 1.0

Sidebars 2
A couple of major regional languages for contrast. Items in bold are not present in the "top 15" wikis. (1) Artikel pilihan, 'selected articles'.

(2) अनुरोध

(3) 'self education'

Articles (millions): id 0.5, hi 0.1

Dark mode
Aron Manning has some dark css mw:User:Aron Manning/Skin themes (perm). His page lists a bunch of other work, both CSS and server-side extensions.

There is also a "mild" light grey theme, and mention of a responsive Vector.

Left sidebar update
Have spent a fair bit of time the last couple of days commenting and !voting at Requests for comment/2020 left sidebar update. Glad I stumbled on that in a timely fashion; can't recall what led me there, might have been one of the Village Pumps.

Related: Reading/Web/Desktop Improvements; Pune Desktop Improvements Study page, PDF (24 user interviews, 1 city in 1 country); Special:Permalink/950222091 on VPI.

Discussion Tools
Editing news 2020 #1 – Discussion tools is at VP(T).

Having them delivered only to User talk:Pelagic/Newsletters isn't working, as I don't check back there often. I'm thinking dual-delivery like with the Signpost would be better. [Update: I was also getting them at meta:User talk:Pelagic, but didn't see the Echo notification.]

I re-experimented with the Reply 1.0 feature at fr:Utilisatrice:Whatamidoing_(WMF)/Brouillon?dtenable=1 and at fr:Discussion utilisateur:Pelagic fr:User talk:Pelagic fr:Discussion utilisateur:Pelagic . Posted at VPT with my inference on how it works. T240640 discusses multiple sig's in a single post, and the decision to treat paragraphs as indivisible units.

Alternate talk markup is discussed at T240640; Anomie mentions WP:LISTGAP.

All the editing things
Got carried away and posted this at T202921


 * 1) “Desktop” web
 * 2) Visual editing
 * 3) Visual Editor / VE / &veaction=edit
 * 4) Source editing
 * 5) New Wikitext Editor / NWE / VE in source mode / 2017 Editor / &veaction=editsource
 * 6) Classic editor(s) / non-Parsoid editors / &action=submit
 * 7) Toolbar / 2010 Editor / enhanced toolbar / box with JavaScript
 * 8) legacy (2006) toolbar gadget / small toolbar
 * 9) No-toolbar / 2003 Editor / plain box without JavaScript
 * 10) (optional CharInsert gadget combined with any of the preceding three)
 * 11) “Mobile” web
 * 12) Visual editing
 * 13) VE with limited toolbar; second page having: edit summary __not__ pre-filled with /* section*/, minor edit checkbox, watch  page checkbox, preview changes button that gets covered by on-screen touch keyboard
 * 14) Source editing
 * 15) (hidden inaccessible form with discrete edit box, edit summary text box pre-populated with /* section */, minor edit checkbox, and watch page checkbox)
 * 16) full-screen wikitext editor; second page having: preview, edit summary box, invisible non-modifiable /* section */ component, and __no__ option to mark edit as minor (even with AMC enabled)
 * 17) talk page editor when not in "read as wiki page" view (similar to or same as App discussion editor, below?)
 * 18) new section experience
 * 19) reply experience
 * 20) mobile App (Wikipedias only)
 * 21) wikitext with syntax highlighting (articles only)
 * 22) discussion editor / line-break-preserving plain-text editor / wikitext-with-no-preview editor (user-talk only)
 * 23) embedded browser, see Mobile web (other namespaces)

TP karma
Not wikipedia-related, but couldn't resist noting this one: https://thenewdaily.com.au/news/state/sa/2020/04/15/toilet-paper-refund/

The other kind of TP
WAID has posted to VPT: Village pump (technical) - to-do: follow-up

Village pump (technical) – thought I'd subscribed to that? to-do: read

Different editor UIs
T202921 and Village pump (technical)

Thursday 16
a feasible approach would be if Flow is run in a separate namespace ("Flow"?); the Talk namespace is left intact, and "watching" an article automatically watches both it's "talk" page and "flow" board. — Arthur Rubin (talk) 01:11, 20 July 2013 (UTC) VisualEditor/RFC

Saturday 11
Depending on your inclination, this is either a fascinating read or dry as dust:

Taxonomy, nomenclature, and Wikidata
I ran into an interesting issue this morning on Wikidata, trying to add Grystes salmoide (Lac. 1802) Cuvier 1829 as a synonym for Micropterus salmoides (Lac. 1802).

First some background. The concept of the largemouth bass as a species seems to have been fairly stable for a couple of centuries. But it has been placed in different genera at different times. These genera are distinct taxa, having different circumscriptions, but the species itself is the same taxon with different binomina. This is unique to species-series names, higher taxa have one-part names that depend on their children (the type taxa) but not parents. (Higher ranks do sometimes have different names for the same scope, but for other reasons.)

Property «synonym» takes an item, not a string as its object, so I need to create a new item for the synonym. This makes sense as biological names have some structure, though you could achieve something similar with qualifiers, and using qualifiers would be reasonable semantically. [add example]

So, whilst "largemouth bass" / Micropterus salmoides is an «instance of» «taxon», I wanted Grystes salmoide to be an «instance of» «scientific name». So far so good. But then I wanted statements for the author and year of the name. I tried «taxon author» and «taxon year» as the closest available concepts; after all, when we state the "taxon author" for a taxon, we're really talking about the first author of the corresponding, not the taxon itself, which may change in extension or intension but keeps the original name. But I ran into a constraint warning: «taxon author» should only be used with subjects that are instances of «taxon» not «scientific name».

So either we have choices:
 * 1) change the definition of «taxon author», etc. to do double duty for both nomina and taxa i.e. alter the constraint
 * 2) create new "nomen author", "nomen year", "nomen rank" properties
 * 3) change the «scientific name» property so that it can only take as its object «scientific name» items, not strings, and change "taxon author" to "nomen author". This separates nomenclature from taxonomy: each «taxon» item has one or more separate «scientific name» items.
 * 4) change «synonym» to take a string, the way «scientific name» does
 * 5) do nothing and just use the broader «creator» and «inception» properties on the «scientific name» item
 * 6) give up and create the nomen as a 'taxon'

Update
I tried (5) and it's still not great. «taxon synonym» wants its object to have a «taxon name», and «taxon name» wants its subject to be an instance of «taxon».

I'm not a fan of repeating the label, auth and year in a «taxon name» element, but it might be necessary to work around the item label being language-tagged: «taxon name» appears to be a language-neutral / multilingual string. Unless we can get people to agree that biological names should be tagged "la". Yes, they are arbitrary Latinate labels and not "Latin language", but so what?

…

I just checked the property proposal for P1420 «taxon synonym». Felix Reimann suggested just using P225 «taxon name» with a lowered status. There was much discussion about making it an item or a string, Brya was cognisant of the taxon-versus-nomen distinction: "Now Wikidata is more or less limited to giving all the accepted taxa their own item (and failing at achieving any kind of minimum quality). With datatype item, Wikidata would be engaged in getting any name ever published its own item. It would become (lots) worse than CoL."

[Sunday, circa 09:00 AEST]

François Péron
Some reading about the expedition of Baudin, Péron, [[Charles-Alexandre Lesueur|Lesueur], & Petit. Unfortunately Leseur's illustrations don't seem to have made it into a free repository that I can find.


 * Péron and the birth of the science of invertebrates, by Dr Gabriel Bittar
 * Charles-Alexandre Lesueur: Biography. Design & Art Australia Online. Sophie G. Ducker, 1992.
 * Terre Napoleon: A History of French Explorations and Projects in Australia by Ernest Scott. 2ed. Methuen, London. Project Gutenberg Australia.
 * Not so much about the scientific aspects; does go into depth about the charts of Freycinet and the imprisonment of Flinders. (added 23:19, 4 April 2020 (UTC))