And breaking this thread to somewhat parallel. I think this might be a good place to post my conversation with la lojbab. that happened not long ago. Sorry for formatting.
Gleki:
> L4B uses an approach of specifying timestamps and dates using either {pi'e} or cmene.
> However, the official definition mentions some {joi} method:
> detri = x1 is the date [day,week,month,year] of event / state x2, at location x3, by calendar x4. (time units in x1 are specified as numbers separated by pi'e or are unit values massified with joi); See also cmavo list de'i, djedi, jeftu, masti, nanca, tcika.
> What is this "unit values massified with joi" method? I can't find its usage anywhere. How it was supposed to work?
Robert LeChevalier:
06/11/15
> I think that is referring to something like "5 djedi joi 11 masti joi 2015 nanca, which of course could be given in any order, whereas the pi'e method requires a standard order in order to know what units the numbers refer to.
> Actually that isn't completely true since someone could used mixed numbers and lerfu separated by pi'e: 5d.11m.2015n and you would probably know the meaning even if the terms were in a different order; but back when we first made the gismu, we had not yet virtually eliminated any difference in grammar between lerfu and namcu.
> There were a lot of people arguing for different orderings for the date components, and we wanted to have a method that could easily specify the ordering for those who wished to use a non-standard order, whatever the standard might have ended up.
Gleki:
>> 5 djedi joi 11 masti joi 2015 nanca
> not sure what you mean. {li 5 djedi joi 11 masti joi 2015 nanca cu detri} obviously doesn't parse.
Robert LeChevalier:
> How about if you just leave off the li? Then you have three quantified sumti joined into one by joi
Gleki:
>You mean {lo djedi be li 5 ku joi lo masti be li 11 ku ...}? But then they can't fill detri1 because detri1 is a number.
>As for {5 djedi} it simply means "5 fullday intervals" that not necessarily go one after another, they can intersect.
Robert LeChevalier:
> We have cmavo that convert *anything* into a number. a sumti, a bridi, whatever. So that would work too, using the Mex conversion cmavo. mo'e can convert a sumti into a number; ni'e can convert a selbri into a number.
> (I have to admit that in our usage, we never said that "detri1 is a number", requiring conversion to a particular kind of construct. Rather, the place structure specifies what NORMALLY goes there, and if something abnormal goes there, one would pragmatically assume the proper type-conversion. Thus pragmatically one would not need the conversion operator, but they are provided in case they are needed for formalism.)
>>As for {5 djedi} it simply means "5 fullday intervals" that not
necessarily go one after another, they can intersect.
>"not necessarily", indeed. But implicit in joi is that they are in some (non-logical) way additively connected. Which way is determined pragmatically, and when we came up with it, this was more or less a conventional pragmatics that made sense.
>An employee who is paid by the hour and works 5 cacra is not going to work 5 overlapping hours, and is unlikely to work 5 completely unrelated hours; the most common pragmatic interpretation is that they are consecutive hours. Similarly with dates - there is no plausible meaning of 5 djedi in a date context that is other-than contiguous and consecutive, because that is what dates measure.
>(We did not put a lot of thought into this question, but largely presumed that all such matters would be resolved by pragmatic convention, which is also what pi'e requires/denotes.)
Gleki:
> May I repost your answers to the mailing list?
Robert LeChevalier:
10/11/15
>May I repost your answers to the mailing list?
Go ahead.