Wikipedia:Bots/Requests for approval/ValhallaBot

The following discussion is an archived debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA. The result of the discussion was

Denied.

ValhallaBot

Operator: Emily Jensen (talk · contribs)

Automatic or Manually assisted: Automatic

Programming language(s): Python

Source code available: Standard pywikipedia replace.py, with exclusion code from Template:Bots/doc#Python

Uses regular expression replacements like

"(\{\{\s*Japanese episode list[^\{\}]*\{\{[^\{\}]*\}\}[^\{\}]*(OriginalAirDate|FirstEngAirDate)\s*\=)\s*(\[\[)?\s*(Jan|January) (\d{1,2})\s*(\]\])?\s*\,?\s*(\[\[)?\s*((19[3-9]|2\d{2})\d)\s*(\]\])?" "\1 {{Start date|\8|1|\5}}"

"(\{\{\s*Japanese episode list[^\{\}]*\{\{[^\{\}]*\}\}[^\{\}]*(OriginalAirDate|FirstEngAirDate)\s*\=)\s*(\[\[)?\s*(Feb|February) (\d{1,2})\s*(\]\])?\s*\,?\s*(\[\[)?\s*((19[3-9]|2\d{2})\d)\s*(\]\])?" "\1 {{Start date|\8|2|\5}}"

etc, seasoned to taste for each template whose uses are being processed.

Function overview: Conversion of linked or raw dates in uses of certain template parameters to use {{Start date}}, if contextually appropriate.

Links to relevant discussions (where appropriate): Wikipedia:BOTREQ#Convert_dates_to_template

Edit period(s): Continuous

Estimated number of pages affected: 10000

Exclusion compliant (Y/N): Yes

Already has a bot flag (Y/N): No

Function details: Date-containing parameters for uses of the templates listed at User:Pigsonthewing/to-do#Date_conversions will be converted from linked or raw format to a form using {{Start date}}, if appropriate for the context in which the date appears. This will allow microformats for the pages using the templates to be encoded correctly.

Discussion

Non-Gregorian dates

Microformats emit dates in the ISO 8601 format, according to the relevant microformat standards. The ISO 8601 standard requires that dates always be in the Gregorian calendar. Some of the infoboxes to be processed, such as {{Infobox royalty}}, are likely to contain dates that are in the Julian calendar, and bots are incapable of discerning whether the date is Gregorian or Julian. I suggest the bot be programed to not process any date before Thursday, 1 March 1923. This is the first day the Gregorian calendar was observed in Greece. It represents the last change from Julian to Gregorian calendar (although some changes from unrelated Asian calendars to Gregorian occurred later). Jc3s5h (talk) 02:14, 27 April 2010 (UTC)[reply]

I've implemented that feature, avoiding modification of dates prior to January 1, 1930, just to be safe. Emily Jensen (talk) 02:24, 27 April 2010 (UTC)[reply]

That's too restrictive; the recommendation on {{Start date}} is to avoid dates outside the range outside the range 1583-9999 CE. Note also, for example, that the infobox on Constantine I of Greece uses the "New Style" (Gregorian) date, per the article's lede. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 09:51, 27 April 2010 (UTC)[reply]

I disagree. The statement about confining dates to the range 1583–9999 is for humans to read, and is motivated in part by the fact that ISO 8601 limits dates to that range unless an agreement among the data exchange partners exists to do otherwise; it isn't clear that such an agreement exists. The full statement in the template documentation is

Therefore, use of this template for non-Gregorian dates or dates outside that range constitutes a false claim of conformance to the ISO 8601 standard.

Since a bot cannot tell from context whether the date is Gregorian or Julian, no date where there is any reasonable possibility of it being Julian should be processed. Pigsonthewing's comment "note also, for example, that the infobox on Constantine I of Greece uses the "New Style" (Gregorian) date, per the article's lede" illustrates the point; the bot cannot parse the article's lede.

Perhaps the bot could generate a list of articles where pre-1923 dates were present so Pigsonthewing can review them manually. Jc3s5h (talk) 11:50, 27 April 2010 (UTC)[reply]

Or perhaps it could produce a list so that Jc3s5h can identify the tiny minority of Julian dates, (if any - no evidence of such dates being use has been offered). The caveat "reasonable possibility" is interesting: There is, for example, no reasonable possibility of Julian dates circa 1920 being found in infoboxes of people from anywhere except Greece. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 12:21, 27 April 2010 (UTC)[reply]

I will not stand for bots that are run with reckless disregard for the truth. Jc3s5h (talk) 12:37, 27 April 2010 (UTC)[reply]

A further requirement is that the bot should detect whether the date was in the format dd month yyyy or month dd, yyyy and use the df parameter so the appearance of the date is unchanged. Jc3s5h (talk) 13:05, 27 April 2010 (UTC)[reply]

Appropriate use of the df parameter will be provided. It appears that there's at least a consensus that dates during and later than 1930 should be converted. The treatment of dates prior to this time is a separate issue. Dates from 1583 to 1929 should be converted only if there is a consensus for this element of the task, which isn't shown by the discussion above. I'll put a notice on WP:VPT asking for additional comments. Emily Jensen (talk) 00:48, 28 April 2010 (UTC)[reply]

There is already bot approval for the task described above. In any case, Jc3s5h's objections do not apply to dates after 1 March 1923; not 1930. Perhaps he would be reassured by some cross-checking with categories, such as omitting anything before 1 March 1923 with a Greek category, and so on? The wider issue of how we label possibly ambiguous non-Gregorian dates (even in prose) perhaps needs to be addressed elsewhere. According to existing policies/ MoS, such dates should already be Gregorian, or clearly labelled otherwise. Is there any evidence that this is not the case? Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 09:04, 28 April 2010 (UTC)[reply]

If you look through the articles on UK monarchs or popes you will find many articles that make no explicit mention of the calendar. Even if there was explicit mention, bots can't parse it.

Greece is only the most recent country to switch. Other 20th century switches include Romania in 1919 and the USSR in 1918. I suggest making a list of articles (and perhaps categories) that contain 1923 and earlier dates, but not processing those dates on the first pass. Review of the list might make certain categories pop out that could be processed safely on a second pass. Jc3s5h (talk) 14:35, 28 April 2010 (UTC)[reply]

If articles make no mention of the calendar used; then the date is either Gregorian, or the example is in breach of existing policies/ MoS, as stated above, and the issue needs be dealt with elsewhere. Bots should certainly be capable of finding strings like "julian", "os"/"o.s." or "old style". Again: is there any evidence of unlabelled non-Gregorian dates in ambiguous years? Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 16:35, 28 April 2010 (UTC)[reply]

The article Guy Fawkes uses dates in the Julian calendar, as one would expect for persons who lived in England in the period from 1583 to 1752. The words "Gregorian", "Julian", and "style" are not present in the article, nor are the abbreviations "O.S." or "N.S."

It is absurd to think that a bot can parse an article to decide if the dates are Julian or Gregorian. Any attempt to do so will trigger a RfC from me.

Furthermore, the style guidelines are just a guideline. Verifiability is a policy, and reckless disregard for the truth violates that policy. Jc3s5h (talk) 17:07, 28 April 2010 (UTC)[reply]

I'm not sure where you're seeing the imaginary "reckless disregard for the truth" you keep referring to. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 18:10, 28 April 2010 (UTC)[reply]

I consider it reckless to think a bot can correctly discern the calender used within articles that transclude infoboxes similar to {{Infobox historical event}}, which is included in this proposal. In the case of the article Maryland Toleration Act the bot would surely get it wrong. Jc3s5h (talk) 17:20, 29 April 2010 (UTC)[reply]

Aside: Until we resolve this, perhaps we can start with templates which have no instances before 1923? These include: various Television series episodes, {{Infobox Aircraft occurrence}}, {{Infobox album}}, {{Infobox concert tour}}, {{Infobox Dotcom company}}, {{Infobox Grand Prix race report}} , {{Infobox OS version}}, {{Infobox shopping mall}}, {{Infobox Single}}, {{Infobox Software}}, {{Infobox space expedition}}, {{Infobox summit}}, {{Spoken Wikipedia}}, etc; plus those from territories where the changeover date was much earlier than the earliest possible use of the template: {{Infobox aus sport club}}, {{Infobox drums corps}}, {{Infobox NFL}}, {{Infobox NHL team}}, {{Infobox NWHL team}}, {{Infobox Pinball}}, {{Infobox Pro hockey team}}, {{infobox UK rail accident}}, {{Infobox Victorian Closed Railway Station}} etc. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 21:05, 28 April 2010 (UTC)[reply]

I have no objection to the boxes listed, but of course "etc." would have to be interpreted with care. Jc3s5h (talk) 17:20, 29 April 2010 (UTC)[reply]

Of course. Thank you. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 22:15, 29 April 2010 (UTC)[reply]

Question

Why is {{Start date}} used when the date isn't necessarily the start of anything? OrangeDog (τ • ε) 11:57, 28 April 2010 (UTC)[reply]

The dates concerned are all the start of something. (For people's births, {{Birth date}} or {{Birth date and age}} should be used.) Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 16:29, 28 April 2010 (UTC)[reply]

"Broadcast date" for an episode isn't a start. {{Infobox NFL}} already has both a {{Start date}} and an {{End date}} (which don't appear to refer to the same thing) as well as multiple other dates. Why can't {{Date}} just be fixed to use microformats (or {{Date mf}} created)? OrangeDog (τ • ε) 17:51, 28 April 2010 (UTC)[reply]

{{Start date}} and other templates were crated and are already used widely and successfully for this purpose;a nd have been for almost three years. This isn't the place to debate alternatives, which in any case don't seem to be needed. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 17:56, 28 April 2010 (UTC)[reply]

Just wondering what the point is in automatically incorrectly tagging dates as "start dates", when they're usually just "dates". OrangeDog (τ • ε) 22:10, 28 April 2010 (UTC)[reply]

While per-article processing is performed automatically, I will make the decision to replace dates appearing as given template parameters with {{Start date}} or other microformat-emitting date templates on a per-parameter basis, and only if contextually-appropriate date templates are available. I've modified the function descriptions to articulate this nuance. Emily Jensen (talk) 00:56, 29 April 2010 (UTC)[reply]

{{Start date}} is always the correct template to use for the requested changes. What others did you have in mind? Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 09:21, 29 April 2010 (UTC)[reply]

{{End date}} or {{Birth date}} come most immediately to mind. Emily Jensen (talk) 16:15, 29 April 2010 (UTC)[reply]

None of the templates in the request are bioggraphical, so {{Birth date}} should not be needed. Some of the templates can take an end date, but that's not included in this request, due to unresolved issues about the use and recognition of inclusive- vs. exclusive- end dates. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 22:15, 29 April 2010 (UTC)[reply]

I've modified the function descriptions to indicate that only {{Start date}} will be used. The contextual appropriateness of the template for each date parameter will be assessed on a case-by-case basis; only dates actually representing commencements will be so encoded. Emily Jensen (talk) 04:00, 30 April 2010 (UTC)[reply]

That's just the job; thank you. Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 12:03, 30 April 2010 (UTC)[reply]

None of the proposed changes will be "incorrect tagging". Andy Mabbett (User:Pigsonthewing); Andy's talk; Andy's edits 09:13, 29 April 2010 (UTC)[reply]

Denied. — The Earwig ^(talk) 20:39, 3 May 2010 (UTC)[reply]

The above discussion is preserved as an archive of the debate. Please do not modify it. To request review of this BRFA, please start a new section at WT:BRFA.