On Fri, 9 Nov 2012, Allin Cottrell wrote: > On Fri, 9 Nov 2012, Sven Schreiber wrote: > >> Am 09.11.2012 18:17, schrieb Allin Cottrell: >>> On Fri, 9 Nov 2012, Sven Schreiber wrote: >>>> >>>> I'm not saying I need it, but it isn't obvious to me why it is less >>>> useful in principle than for example 'isstring()'. Instead it seems to >>>> be a natural complement to the other 'is...()' functions, doesn't it? >>> >>> Maybe I was too glib. But actually, it seems to me that all we >>> really need is isnull() -- >> >> [followed by more detail] >> >> Hm, not sure, but what about the case of analyzing the components of >> bundles in a script? Maybe the bundle itself is a function argument, and >> then the function needs to check the sanity of the structure of the >> bundle that was passed in? > > Point taken. I need to think about that. If it's valid I guess > we should add ismatrix and isbundle? Or, perhaps better, roll > these distinct functions into something we might call typeof? > > if typeof(whatever) != "scalar" > ... > endif
I like this. Moreover: sometimes we run into trouble because automatic re-casting of a scalar as a 1x1 matrix doesn't work. For example: <hansl> scalar c = 5 matrix x = invpd(c) </hansl> now works, but it wouldn't not so long ago (invpd would complain its argument should have been a matrix). Of course the above not working was a bug, and was killed as such, but sometimes it's nice to have a workaround while the bug is still there. The only think I'd change from Allin's proposal is to introduce an integer-based coding system, such as 1=scalar, 2=series etc and have typeof() return a scalar. For performance reasons, the less we use strings, the better. -------------------------------------------------- Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche (formerly known as Università di Ancona) r.lucchetti(a)univpm.it http://www2.econ.univpm.it/servizi/hpp/lucchetti --------------------------------------------------
