Re: [Gretl-users] declare a list to a bundle

On Thu, 18 Aug 2016, Artur T. wrote: > Hi all, > > I would like to declare a list "z" to a bundle "b" but I am surprised to > see that this doesn't seem to work: > > <gretl> > open denmark.gdt --quiet > bundle b = null > list z = const LRM > b.z = z > </gretl> > Message: "The variable b is of type bundle, not acceptable in context" Actually, that message is kind of backwards; it's really the list z that is "not acceptable" in the context of adding something to a bundle. We'll see what we can do about that. > Is this actually intended? Also, is there in alternative how to handle > this (I disregard the possibility to define a matrix which could be send > to "b" for the moment). Yes, it's intended. The point is that bundles are not tied to, or synchronized with, the life-cycle and mutations of a dataset. A bundle that was created with dataset D1 in place may be read at some point when D1 has been extensively modifed, or replaced by a different dataset D2, or when there's no dataset in place at all. Since a list as such is a record of the ID numbers of a set of series in a given dataset, it doesn't make sense (in our view) to allow putting a list into a bundle. As for the matrix-based alternative, note that there are two meanings for "creating a matrix out of a list": matrix m = {list} creates a matrix which contains all the data referenced by list, while matrix m = list creates a vector holding the ID numbers in list. You can use the latter method to "emulate" putting a list into a bundle. Then when you want to extract it you can say [catch] list L = b.m to turn it back into a list. The virtue of this approach is that it will automatically flag an error if m cannot be converted to a list given the current dataset (or absence of one) -- that is, if it contains any elements that are out-of-bounds if considered as ID numbers of series. It's implicit in the User's Guide that lists can't go into bundles (lists are not mentioned as a valid content type) but it would be a good idea to make this explicit, and present the matrix-based "workaround" for those who really want to bundle lists. Allin _______________________________________________ Gretl-users mailing list Gretl-users(a)lists.wfu.edu http://lists.wfu.edu/mailman/listinfo/gretl-users