Well, even presuming that (data) typing is implicitly specified by the CLL and various definitions (such as on jbovlaste), there are some questions that I have.
First, is it reasonable to inherently impose inherited/syntactically-specified type restrictions on the empty/universal argument based on syntactic context? For example, if lo lerfu can only be a BY construct (by type), "zai'o lerfu" should reasonably have zai'o mean only the empty lerfu (as opposed to, say, the empty set). The empty/null objects of different types are, well, different and the type requirements should still be respected, I opine.
Second, I figure that the type specification should generally be maximal. However, it could be useful to restrict that type specification further. For example, lo selsku can be any of the following (types): sedu'u/text/lu'e concept; so, in "cusku zai'o", zai'o (by maximality) refers to the empty sentence-abstraction and/or the empty text (empty string?) and/or the empty symbol (id est: refers to the empty object associated with the union of the aforementioned types); if only the empty symbol is desired to be referenced, then restricting the type of zai'o in this context would be necessary. How would one go about restricting the type? Is a simple poi clause good enough, or would a metalinguistic restrictive clause be necessary?
Third, continuing from the second point, this definition allows for maximality of potential type only within the range of syntactically permitted types. Could it ever be useful to remove this restriction and allow the referent to potentially be the empty object of any possible type? For example, this possibility could allow for "1+∅" arising. This example does not seem to be very elucidating or useful to me, but I am not sure that there generally exists no such example. If this option is to be allowed, how can we preserve the (probably more) useful functionality of maximal genericness within allowed types as presented in the current definition while enabling this new functionality? (How do we unrestrict the typing?)
|
-
Comment #2:
Re: Typing
|
gleki (Sat Jun 14 13:47:19 2014)
|
krtisfranks wrote: > trimmed...
I'm not a progger. Is zai'o for number/boolean = null, zai'o for string = ""?
Looks like the concept of sumti types has been mostly diassembled by xorxes in Simple gimste revision ("text" and "sound" might be a subtype of "object/event" althought I'm retaining them for pragmatics purposes).
However, what cannot be shut down are tersumti interactions (cpedu2 clearly contains a hard link to cpedu3) and plurality.
|
-
Comment #3:
Re: Typing
|
Curtis W Franks (Sat Jun 14 16:53:39 2014)
|
gleki wrote: > krtisfranks wrote: > > trimmed... > > I'm not a progger. Is zai'o for number/boolean = null, zai'o for string > = ""? Yes, I do think so. And even for language, it could possibly be the empty language (although that assumes a few things about the languages being considered and requires that language is being thought of as a type).
> Looks like the concept of sumti types has been mostly diassembled by > xorxes in Simple gimste revision ("text" and "sound" might be a subtype of > "object/event" althought I'm retaining them for pragmatics purposes). I am not sure what you mean. But I do think that some terbri specify (at least some of the acceptable) types in their official definitions. If typing is not really a thing in Lojban, or is done only by individual interpretation, then pretty much all of the mentions and discussion about type can be ignored and this word can be of as general of a type as desired/reasonable. I reälize that the whole topic of types is a big one; a more in-depth conversation should be conducted elsewhere; I was just trying to address the issue in case it came up. (If types are ignored, either we have to come up with a whole bunch of cmavo for each empty/null object (of each type) or we have to allow for type specification/restriction by possibly wordy and inconvenient means in basically every situation)
> However, what cannot be shut down are tersumti interactions (cpedu2 > clearly contains a hard link to cpedu3) and plurality. How do you think that these interactions can affect the reference of this word?
|
-
|
|
|