Proposal: loi lerfu tcita detri; the final word on the problem of dates and times?: Difference between revisions

From Lojban
Jump to navigation Jump to search
No edit summary
No edit summary
Line 65: Line 65:
Additional crazy idea: What if we also allowed periods-of-time? Like in ISO 8601, this would be accomplished by placing a special prefix right at the start of the '''detri'''. For example: '''ze'a lo se detri be li ky ny ci dy vo''' could mean "for three years and four days"... ('''ky''' standing for '''kuspe'''? '''ditcu''' definitely can't be used, '''temci''' might be too close). This does kinda stretch the meaning of '''detri''' though; perhaps another cmavo and/or brivla can be proposed.
Additional crazy idea: What if we also allowed periods-of-time? Like in ISO 8601, this would be accomplished by placing a special prefix right at the start of the '''detri'''. For example: '''ze'a lo se detri be li ky ny ci dy vo''' could mean "for three years and four days"... ('''ky''' standing for '''kuspe'''? '''ditcu''' definitely can't be used, '''temci''' might be too close). This does kinda stretch the meaning of '''detri''' though; perhaps another cmavo and/or brivla can be proposed.


Additional problem: One thing that comes up is, how to refer to the numbers contained within these tags. Here are some ideas (let's say '''ko'a''' refers to the date):
Additional problem: One thing that comes up is, how to refer to the numbers contained within these dates. Here are some ideas (let's say '''ko'a''' refers to the date):
* {{jvs|etmoi}}: '''lo etmoi be me'o cy bei ko'a''' would then refer to the value tagged by '''cy''' (hour of day).
* {{jvs|etmoi}}: '''lo etmoi be me'o cy bei ko'a''' would then refer to the value tagged by '''cy''' (hour of day).
* Given the usage of {{jvs|memimoi}} to mean "mine", it could be argued that '''moi''' <u>is</u> another way to say '''etmoi'''. Thus, '''lo me me'o cy moi be ko'a''' could work. (The '''me me'o''' is required though, as bare '''cy''' refers to the variable-pronoun that bare Lojban letter-cmavo stands for, not the letter itself)
* Given the usage of {{jvs|memimoi}} to mean "mine", it could be argued that '''moi''' <u>is</u> another way to say '''etmoi'''. Thus, '''lo me me'o cy moi be ko'a''' could work. (The '''me me'o''' is required though, as bare '''cy''' refers to the variable-pronoun that bare Lojban letter-cmavo stands for, not the letter itself)

Revision as of 17:10, 5 August 2015

Initially proposed by la zipcpi.

Lujvo name: lertcitydetri

Note: The experimental parser grammar currently separates letter-strings from number-strings (in accordance with another proposal with the aim of reducing necessary use of boi), breaking the use of this system. The experimental cmavo li'ei is suggested in place of li/me'o to maintain compatibility.

There has been much debate about how the li-number in the x1 place of detri (and consequently, the sumti of the preposition-tag de'i) should be represented. As far as I can tell, there are two major camps:

One camp wants to use DD-MM-YYYY, separated by pi'e. Lojban for Beginners also promotes this usage. Its advantages are that people often refer to the date or the date and month without specifying the year (whether because it is obvious from context, or because it refers to an event that happens every year); thus there is less need for empty pi'es. However, this has several disadvantages:

  • Possible confusion for people who are used to MM-DD-YYYY systems.
  • Specifying years will still need empty pi'es.

The other camp wants to follow the ISO 8601 system, also known as YYYY-MM-DD. This adopts the advantages that the ISO 8601 system has; namely, international applicability (less likely to confuse people who normally use either DD-MM-YYYY or MM-DD-YYYY), and having prescribed formats for many alternative countings like week-of-the-year and nth-day-of-the-year. However, it has one huge disadvantage: referring to the date or month while eliding the year requires use of empty pi'es. Also, pop quiz: if you want to refer to the 25th of the month under ISO 8601, how many pi'es should you use?

Did you say two? Because you would be wrong. The correct answer is three. The ISO 8601 specification says that a single dash is used to truncate the century. Two dashes would truncate the year. Three will truncate the year and month, leaving the day. (Seriously, it's right there on page 12 on the above linked document.) The problems are obvious; even if both the speaker and listener are aware of this pitfall, will they be able to count the pi'es accurately? For all its advantages, the ISO 8601 was just not developed with Lojban in mind; certainly no English speaker is going "dash dash dash twenty-five" when talking about dates.

Additionally, the two systems directly clash, causing possible ambiguity, especially when talking about dates in the early years of the AD/CE epoch.

Hopefully no serious Lojbanist is suggesting that we use MM-DD-YYYY! zo'o

Anyway, because of all these problems, and after some discussion in the IRC chatrooms; I propose, loi lerfu tcita detri ("letter-tagged dates"), also known as the "NLDCMS system": a unified Lojbanic system for dates and times.

Essentially, they look like this; 17 May 2015, 22:05:23 will be: (detri fa / de'i) li ny renopamu ly mu dy paze cy rere my mu sy reci or, to shorten the letters and numbers, de'i li N 2015 L 5 D 17 C 22 M 5 S 23 (alternatively, me'o can be used instead of li.)

Essentially, each part of the date has a letter-tag associated with it. lo ve detri is considered to be a computer-program-like-system that could take all these weird mixed-base strings of letters and numbers and convert it to a point or duration in time. The letter-tags are:

  • ny : year, for nanca
  • ly : month, for lunra (the proper Lojban word is masti, but this metaphor is used to avoid a clash with mentu minute)
  • dy : day, for djedi
  • cy : hour, for cacra . This will use the 24-hour system.
  • my : minute, for mentu
  • sy : second, for snidu

Additionally, there are additional tags for the additional use-cases covered by the ISO 8601:

  • nydy : day-of-the-year. For example de'i li ny renopamu nydy remuxa refers to 13 September, 2015, the 256th day of the year, Programmer's Day.
  • jy : week-of-the-year, for jeftu (as defined by the ISO week date)
  • jydy : day-of-the-week, starting on Monday; this is assumed for the dy place iff the jy place is defined.
    • zy: proposed single-letter variant of jydy (as days of the week are often referred to), with mnemonic ze
  • jyny : ISO week year, where it is necessary to distinguish it from the Gregorian year; alternatively an empty jy place can be placed after the ny place
  • vy : time zone, for ve tcika. This would be given in terms of hours after UTC (vy ni'u = hours before UTC). So de'i li cy pavo my no vy ni'umu means "14:00, UTC-5" . Timezones that also have minute-offsets will use pi'e, e.g. de'i li cy ze my cino vy mupi'evomu means "07:30, UTC+5:45". To refer to a specific location instead of an offset, use pe bu'u [ko'a], te de'i [ko'a] or ve ti'u [ko'a] instead, e.g. de'i li cy pano my no ve ti'u la .tokiios. = "10:00, Tokyo time". (ti'u / tcika may be preferred instead when specifically talking about the time of day; however, this system should work for both detri and tcika.)
  • I assume that eliding the century won't be common, but it can be accomplished for translation purposes by using xo'e: de'i li ny xo'epavo = "the year '14".
  • xy : century, for xecto; de'i li xy repa = "the 21st century"
  • In audio communications with possibly-compromised sound quality, the tags can be made easier to tell apart via word+bu, for example nanca bu

"Back-compatibility" tags for 12-hour time; discouraged unless absolutely necessary:

  • cycy : a.m., from clira cacra / cerni cacra
  • lycy / vycy : p.m., from lerci cacra / vanci cacra
  • zycy : unknown/unspecified 12 hour time; zy from zo'e

The advantage is that by having unique letter-tags for each part of the date, any element can be left out (or even reordered). For example, to say "the year 2005", it is de'i li ny renonomu; to say "the 31st of the month", it is de'i li dy cipa; to say "December 25th", it is de'i li ly pare dy remu . The letter-tags also act as a semantic anchor, preventing one from losing count as to what each number is supposed to refer to. Additionally, the letter-tags uniquely identify this system, thus it is unlikely to be confused with any previous usage of dates.

This system is also possibly extensible to other ve detri/calendar systems; however, someone familiar with those systems will have to figure out how this is to be accomplished.

Additional note: It's also sometimes useful to turn a date into a sumti. (Things like "every December the 25th") Here are some methods to do this:

  • ro lo se detri be li ly pare dy remu: accepted by official grammar.
  • ro la'e de'i li ly pare dy remu: requires a proposed extension of LAhE to cast termsets into sumti, originally made to allow mi jinvi tu'a naku to mean "I think not" (mi jinvi lo du'u na co'e, with co'e being referential.)

Additional crazy idea: What if we also allowed periods-of-time? Like in ISO 8601, this would be accomplished by placing a special prefix right at the start of the detri. For example: ze'a lo se detri be li ky ny ci dy vo could mean "for three years and four days"... (ky standing for kuspe? ditcu definitely can't be used, temci might be too close). This does kinda stretch the meaning of detri though; perhaps another cmavo and/or brivla can be proposed.

Additional problem: One thing that comes up is, how to refer to the numbers contained within these dates. Here are some ideas (let's say ko'a refers to the date):

  • etmoi: lo etmoi be me'o cy bei ko'a would then refer to the value tagged by cy (hour of day).
  • Given the usage of memimoi to mean "mine", it could be argued that moi is another way to say etmoi. Thus, lo me me'o cy moi be ko'a could work. (The me me'o is required though, as bare cy refers to the variable-pronoun that bare Lojban letter-cmavo stands for, not the letter itself)