dates

From Lojban
Revision as of 08:13, 30 June 2014 by Conversion script (talk) (Conversion script moved page Dates to dates: Converting page titles to lowercase)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to: navigation, search

A convention for dates

There are those who prefer big-endian for dates (year, month, day), according to ISO 8601. They say it's easier to sort, and want to adhere to international standards. Then there are those who say that little-endian (day, month, year) is the way to go because we discuss days more than we years. There is an elegant solution to satisfy both camps.

We should reverse the direction of elision of pi'e pieces.

2001;9;17 (full date)

2001;; (only the year)

9; (only the month)

17 (just the day)

--xod


Given that Nick and I (rab.spir) represent the big-endian camp, I think you've more repulsed us than anything. This middle-endian convention is bizarre and unusable. In fact, if you insist on proposing any more for this "compromise", consider me switched to the little-endian side for the sake of sanity.

  • Well Jim Carter and I like big-endian as well. However, if this proposal settles the debate by scaring the big-endians into little-endianism, that is a strange sort of success too. It's a little like choosing to drive on the right or the left side of the road. There may be arguments for either, but the most important fact is that the community picks a direction and sticks to it without deviation. --xod

This seems very convincing! Hence, does it mean:

pi'esopi'e (;09;) (only the month)? Or, better:

sopi'e (09;) (just the month)?

BTW, shouldn't {pi'e} be better transcribed by ":" rather than by ";"? --.aulun.

  • I am not convinced at all. Adding pi'e onto the end should never change the value of a number - not to mention the month problem. Even day-month-year is better than this. --rab.spir There are people who refuse to use day-month-year.
  • When I write "5 inches" or "5in.", the part at the end signals the "value", as you say, of the number. So "year" is signified by "pi'epi'e" and "month" by "pi'e". I don't see what's so bizarre about it and in fact the more I think about, the more I like it. --xod
    • The problem I see here is that it gardenpathing ardenpaths. Someone starts talking about a date, and says "li ci" - okay, so it's the 3rd of the month that they're talking about. No wait, then they say "pi'emu" - now it's March 5th. Then they say "pi'epapa" - now it's May 11, '03. In big-endian dates, what you've already said retains its meaning. --rab.spirAnd when you say "5 inches" you're specifying the unit. "pi'epi'e" is not a unit, and in fact its meaning changes with context. The equivalent for dates would be using a date/time bridi.
    • A "gardenpath" of one word, or within such a small context of several numbers, can't be subject to the same restrictions as one of noticeable size. Although "inches" is a unit, "5 inches" is still subject to "gardenpathing" of the same sort, just like "5 thousand". How about "5 past 4"? Each word alters the meaning of the "5", but nobody finds this usage difficult. "pi'epi'e" will function as a unit; it will always mean "year". If a date is being given incredibly slowly, and the listener is supposed to be able to process each fragment before completion (an unlikely pair of conditions), then this is a problem. In reality, with printed text scanned by the eye, and with speech uttered at competent speeds, I do not forsee confusion occurring. The best solution, rather than scrapping this backwards-elision method, is to provide users with an alternative to it which is formal, not purely numeric, big-endian, and parses slowly, quite like your bridi idea. Then we can let usage decide! --xod

According to the Keyboard key names age of special character names, ";" is pi'ebu and ":" is zo'ubu. But the Book shows ":" for pi'e. I would personally prefer ":" for pi'e and ";" for zo'u! --xod

Quite clearly it doesn't matter. If you don't say the actual cmavo, the reader is going to have to glork anyway. If you don't want the reader to glork, say the word.

sopi'e for month? Absolutely not. The last thing we need with a clash of big-endian and little-endian is middle-endian. Who's to say sopi'e isn't day, followed by omitted hour? I'll thank you to keep that pi'e right there where it belongs: pi'eso. -- nitcion, who's still not sure about this.

Nope. If ";9" means month, then day must be ";;17", which means the most-used number bears the most encumbrance.

  • I don't know who's saying this, but this is moronic. ";9" = month commits you to neither big-endian nor little-endian. You can say ";9" is September and "17;;" = "17" is still the 17th of it; that's in fact the current official system. It does commit you away from middle-endian, and I find it flabbergasting middle-endian is even being entertained. The value that fills the blank field has actually not been specified, and can vary between no'o and tu'o according to endedness. -- nitcion
  • You are not being clear, Nick. Are you suggesting that ";1" should mean January, but "1;;" should mean the 1st day? And secondly, I resent your attempt at a derogatory term with the silly use of "middle-endian". There's clearly nothing middle about it! It's big-endian with pi'e elision in a direction that actually reflects the fact that people discuss days more than they discuss years. There is, after all, a ratio of several hundred to one between them. Do you REALLY want the word for "day" to be four syllables long? Do you think the rest of us do too? After 10 minutes of conversation planning a road trip my mouth will elide the entire "pee heh pee heh" and rely on context and cooperative listeners to figure out what the heck the numbers mean. --xod
    • ( *smile* here we go) I'm suggesting that, whatever the direction of elision, the pi'es have to be left along, as placeholders, to avoid ambiguity. Thus: ";1;" January can be rendered as ";1", but not as "1;" --- that's needlessly confusing. Whether "1;;" means 1 A.D. or the 1st day of the month is completely up to whether you're being big-endian or little-endian. But "1;" is middle-endian, because if a pi'e number is a number like any other (and it is), then whatever direction you're eliding digits from, the first number you see is the "end".That's one. Two, what I know is that in Greek, we don't say "3" for the third of the month. (Hell, you don't in English either.) In fact, we don't even say "on the Third" --- because the word for "Third" already means "Tuesday" --- or "on Three" --- because that already means "3 o'clock". We say "stis tris tu minos": "On the 3 of the month." Do I think "peeheh peeheh" is a reasonable thing for you to put up with? Absolutely. Greek does a near equivalent, and survives; why should I think English got this right and Greek didn't, when English does even say "three" but "the third"? But I have no objection to you saying de'i li ci or de'i li ci pi'e pi'e. Do that to your heart's content; it's clear you're not talking about a year, so you're not really breaking anything. I do object to de'i pa pi'e. The one thing Big-endian and Little-endian agree on is that January is de'i pi'e pa. You're breaking this --- and you're only breaking this now, you'd left it alone in your original proposal. I can only repeat what Rob said: if you make "ny pi'e" differ from "ny", you are breaking the fact that "ny" is a number. And as part of the bargain, you're treating pi'e like "inches" --- which breaks its established definition totally. I'd rather the status quo of little-endianness, and cmene for years, than that. I know you're trying to cut the Gordian knot with this, but you're cutting too much. I cannot accept this as a compromise. (I'm reminded of the two restorations of Cornish competing for speakers. Some dude decides to come up with a compromise scheme combining both. Sure enough, the end result was not one restoration of Cornish --- but three. :-)
  • Show me all this supposed usage of the day alone. Start using Lojban conversationally and you'll see that days ("I come back from Hawaii on the 3rd") are more often used than years.
    • That's neither an answer (Lojban could easily force you to say "3rd of the 7th" --- English, in fact, says 7th, not 7, but 2001, not 2001th, so guess which number it takes as unmarked), nor counterevidence ("it will because I say so" --- your ideology will obviously guide what you use "conversationally"), nor culturally neutral (where I come from, people say "next Tuesday" more often than "the 3rd"; days of the month are an artefact of Rolodex calendars.) -- nitcion
    • Do you mean to tell me that in your universe, Nick, people discuss years more often than they discuss days of the month, in daily conversation? Again, you're not being clear. --xod
    • Nick: Both you and Robin say that you discuss numerical dates more than years, and I'm sorry, but I don't buy it.
      • You're telling me that you refuse to accept my stated description of how I live my actual life? That's amazingly insulting, you know that, right? I assure you, with my social life I end up in conversations like, "Well, I'm booked the evening of the 26th. You?" "Hmmm. 27th?" "Nope. How about Saturday the 29th?" "Yeah, I suppose that'll work." "OK then". I would say, on average, I discuss the day of the month perhaps 20 times more often then I discuss the year. No, I'm not kidding at all. I sometimes forget what year it is; it's that uncommon.
    • Nick: This may well be a cultural thing, but again, in my Universe (otherwise known as Greece), people speak of Tuesdays much more than 17ths. I hear "the 17th", I think "Rolodex", not "humble folk of the soil." (Greece is too small for road trips to be a useful counterexample, too. And after 10 minutes of conversation talking about the history of the Panhellenic Socialist Movement, my mouth will elide the entire "pee heh pee heh" just as well: two can play at that game...) And again, the unmarked number in natural language is consistently the year, not the day of the month; it's the day of the month that has the extra ordinal hanging off it.That said, again, I could learn to live with 2000;;, annoying though it is (pi'epi'e isn't more annoying lengthening a year expression from 5 to 9 syllables than a day expression from 3 to 7? I think 5 vs. 7 syllables for years and dates (li renonono, li pi'epi'e revo) a less grievous imbalance than 9 vs. 3 (li renonono pi'e'pie, li revo).) But I don't want to deal with 1; instead of ;1(;). Not a number any more.Taking a deep breath and stepping back, I think what you've actually achieved with your emended proposal is a instance of the atismo effect: people are now going to avoid numerical expressions for months like crazy, and use cmene instead. Kind of a shame, since ";n;" is unambiguous...

As for times, if you want to glob dates and times onto the same field, you must suffer a year written as 2001;;;;; or an hour written as ;;;20, depending on which direction you choose to elide. Remember that ";;;" has six syllables. This is a strong argument for using separate date and time fields. But when a full date + time is given (2001;9;11;10;40;), all those pi'es will leave no question as to the meaning. --xod


.i tugni mi'e .aulun.


I'm starting to feel that numbers and decimal points cannot convey a date by themselves. I suggest this: date/time bridi --rab.spir

Digits and punctuation are just fine in every culture that uses computers or writes checks. They simply disagree on the proper order.

Good point. I wrote that in a bout of frustration with this whole debate. We still do need a purely numerical way to express dates and times. --rab.spir


Counterproposal, that actually really will satisfy all camps. :-)

All years must always be four digit. If you use two-digit years, you deserve to be pilloried. If you are talking about 1st century BC or AD, for God's sake say which endianness you're using up front.

If you're being Big-endian:

Year 2000

Month  ;9

Day  ;;24

Year Month 2000;9

Month Day  ;9;24

Year Month Day 2000;9;24

If you're being Little-endian:

Year  ;;2000

Month ;9

Day 24

Year Month  ;9;2000

Month Day 24;9

Year Month Day 24;9;2000

If you keep years four-digit, there is no reason there should ever be real ambiguity. Month is unambiguous, date is unambiguous, you can actually freely alternate between Big and Little with impunity, and pi'e remains an analogue of pi. I can see the choice between the two being stylistic, in fact: Big-endian for history, little-endian for travelogues...

-- nitcion

  • I'm down with that. I definately like it better than xod's proposal. -RobinLeePowell
  • Me too. If everyone saying years uses all four digits (de'i li renonore), there's never any ambiguity and thus it's quite fine. For talking about the first century, how about explicitly saying those 0 places (nonoreci == 23, etc). --mi'e .djorden.
    • .i .ua .i'ese'inai certu .i .o'ose'i mi pu na pensi la'ede'u -- mi'e nitcion (Bingo. Why didn't I think of that. No, obligatory four-digit years makes all the sense in the world, and to not have thought of it shows me up as being bound by conventionalities...) As a practicality, though, and to minimise deviance from the Western norm, could I have three digit as well as four digit numbers allowed for years? A three digit decimal number will never be confused with a day or month --- unless we have day/year dating, and if we do, we will invoke a new rulebook. That way, the year 490 BC is ni'u vosono , and the year 70 AD is ma'u nozeno

What mystifies me beyond belief is the perception (I would never accuse a reasonable person like you of actually suggesting this!) that the following notations are acceptable to you:

  • 2001;;
  •  ;;2001
  • 1;;
  •  ;;1
  •  ;9
  •  ;9;

While this is unacceptable to you:

  • 9;

I'm sure I didn't understand something.

  • xod, the notation is not what I have a problem with, but the meaning. To me, pi'e is still a numerical placeholder. Therefore, I accept 9; = 9, but not ;9 = 9;. I guess I'm prepared to accept elision from the left of digits, but not of pi'e placeholders. Your solution is elegant, I fully admit that (in fact, I really like Little-endian reversed, the way you're presenting it below)... but it breaks pi'e, because it means that pi'e is no longer analogous to pi in dates --- though it will remain analogous to pi in, say, base 20 numeration. Date use of pi'e and non-date use would then become incommensurable.Can we do yet another compromise, and call this backwards eliding pi'e of yours an experimental cmavo, like pi'ei?
    • Since pi'e is intended for non-standard bases, I don't see why it can't work for a backward elision if it can work for a negative base like any little-endian scheme. However I have my own reason for preferring an x-cmavo to pi'e: pi'e has two syllables and I want this to have only one. Now the only problem is that this was offered as a solution to the desire for little-endianism, yet now you want little-endian with reversed elision, which for me is just as bad as normal little-endian because of the extra load on days, and the years getting off scot-free. --xod
      • For the x-cmavo, I'd suggest pie. I said I like little-endian with reversed elision; not that I want it. And I do want the years scot free, that's still true; I've argued for it enough already. -- nitcion.

Secondly, I don't believe you've given a convincing reason why the pi'e elision could not be reversed in both of your very clear examples. The result would be like this:

If you're being Little-endian with reversed-elision:

Year 2000

Month 9;

Day 24;;

Year Month 9;2000

Month Day 24;9;

Year Month Day 24;9;2000

If you're being Big-endian with reversed-elision:

Year 2000;;

Month 9;

Day 24

Year Month 2000;9;

Month Day 9;24

Year Month Day 2000;9;24

--xod

  • xod, that's just silly. All your examples have the same number of syllables as Nick's, but have backtracking where his don't, and require more thought. --RobinLeePowell
    • Whatever. I haven't thought about this stuff in months. But I do know that I already addressed the backtracking issue in sufficient length, if you'd please read the above content... --la xod

Shouldn't dates and times have the same endianess? -travys


Since pi'e (at least suggestively) means "has part", it makes sense to always have the bigger thing to the left. The convention that years have four (or three) digits and that months and days have two (or one), will give the following nice table:

Year: 2000

Month: use the month name (how often do you write just the month numerically without context?)

Day: 24

Year Month: 2000;9

Month Day: 9;24

Year Month Day 2000;9;24

I really can't see the need for being "little-endian", this will only cause confusion, and to reverse the meaning of pi'e just for this special case (chapter 18 of the CLL implies strongly that the bigger part goes to the left) seems unwarranted.

mi'e LaNorpan