Re: [Gretl-devel] more fun with bundle members

On Tue, 12 May 2015, Sven Schreiber wrote: It's necessary to distinguish the two parsers in gretl: the one for commands, and the one for assigment + functions calls. In regard to the latter (that is, in constructing function arguments or the right-hand side of an assignment statement) you're supposed to be able to use any valid expression, with arbitrary nesting. E.g., <hansl status="valid"> bundle b b.x = 0.1 scalar y = 16*exp(sqrt(b.x*3^5-8)) </hansl> And if that fails, it's a bug, no question about it. In regard to commands, the acceptable input is much more limited. Many commands (including "var") specify that parameters such as order "may be given numerically, or as the name of a pre-existing scalar variable". So, for example, like your example above, the following will fail <hansl status="invalid"> scalar p = 6 var p/3 IBO IDE </hansl> Now admittedly one could argue that giving the name of a bundle plus a key that identifies a scalar member ought to count as giving "the name of a pre-existing scalar variable", but it's not literally true and it would take a good deal of work to make it so; and one could apply the same argument to a single matrix element, even if identified in some round-about way... leading in the limit to any expression that identifies a scalar. In gretl, "commands" are nice and simple, and you pay a price for that. Allin