EDTF level 2 unspecified digit pattern displays nonsensical

[kspurgin]

Currently:

156X-12-25 -> unspecified year in December 25, 1560
15XX-12-25 -> unspecified decade in December 25, 1500
XXXX-12-XX -> unspecified year and day in December , 0000
1XXX-XX -> unspecified century and month in 1000
1XXX-12 -> unspecified century in December , 1000

Possible improvements:

156X-12-25 -> December 25 of unspecified year in range 1560 to 1569
15XX-12-25 -> December 25 of unspecified year in range 1500 to 1599
XXXX-12-XX -> December of unspecified year
1XXX-XX -> Unspecified month of unspecified year in range 1000 to 1999
1XXX-12 -> December of unspecified year in range 1000 to 1999

[kspurgin]

meta-issue: #1748

[seth-shaw-unlv]

Looking at this again, depending on how you setup the formatter settings, anything we do here would be nonsensical. These examples rely on the formatting to be middle-endian with the full month displayed. However, the default resembles the ISO format (big-endian with hyphen separators), which results in results like:

unspecified year in 1560-12-25
unspecified decade in 1500-12-25
unspecified year and day in 0000-12
unspecified century and month in 1000
unspecified century in 1000-12

I started working on the suggested improvements, but with the default settings, they don't look much better:

unspecified year in range 1560 to 1569-12-25
unspecified year in range 1500 to 1599-12-25
unspecified day in unspecified year-12
unspecified month in unspecified year in range 1000 to 1999
unspecified year in range 1000 to 1999-12

So, I'm not sure what would make sense with these varying configuration options.

[seth-shaw-unlv]

I thought I would also check how the Professional-Wiki EDTF library would handle this:

>>> $parser = \EDTF\EdtfFactory::newParser();
=> EDTF\PackagePrivate\SaneParser {#4667}
>>> $humanizer = \EDTF\EdtfFactory::newHumanizerForLanguage( 'en' );
=> EDTF\PackagePrivate\Humanizer\InternationalizedHumanizer {#1716}
>>>
>>> $humanizer->humanize($parser->parse('156X-12-25')->getEdtfValue());
=> "December 25th, 1560s"
>>> $humanizer->humanize($parser->parse('15XX-12-25')->getEdtfValue());
=> "December 25th, 1500s"
>>> $humanizer->humanize($parser->parse('XXXX-12-XX')->getEdtfValue());
=> "December"
>>> $humanizer->humanize($parser->parse('1XXX-XX')->getEdtfValue());
=> "1000"
>>> $humanizer->humanize($parser->parse('1XXX-12')->getEdtfValue());
=> "December 1000"

Using their humanizer would improve internationalization (as we could instantiate the humanizer for the currently enabled language) while losing all configurability; but at least we could shift finger pointing to them. 😉

[kspurgin]

>>> $humanizer->humanize($parser->parse('156X-12-25')->getEdtfValue());
=> "December 25th, 1560s"
>>> $humanizer->humanize($parser->parse('15XX-12-25')->getEdtfValue());
=> "December 25th, 1500s"
>>> $humanizer->humanize($parser->parse('XXXX-12-XX')->getEdtfValue());
=> "December"

The above are better than what it's currently doing, for sure.

>>> $humanizer->humanize($parser->parse('1XXX-XX')->getEdtfValue());
=> "1000"
>>> $humanizer->humanize($parser->parse('1XXX-12')->getEdtfValue());
=> "December 1000"

I guess they decided that two zeros (century) was the limit for handling with 's. Well, they are not worse than what we have now.

I was not around for discussions of needs/use cases around all the flexible configurability of the EDTF field, so not sure what to say about that tradeoff.

If we were handling it in the current EDTF field, I was imagining the algorithm would have to go something like the following for 156X-12-25 -> December 25 of unspecified year in range 1560 to 1569:

• augh this value has X's in it, I have to do a special thing which will maybe override more normal display settings
• year part has X(s) so I turn it into a range for display (unspecified year in range 1560 to 1569)
• I have month and day without X(s) so I will turn that into (December 25 or 25 Dec or whatever I'd do with those based on display settings)
• Concatenate the known part with the range part, separating with of

Date stuff is SO FUN. :-D