Re: [Gretl-devel] Variable name clashes with function name
Am 22.12.2018 um 14:10 schrieb Allin Cottrell:
On Sat, 22 Dec 2018, Berend Hasselman wrote:
> When I run the standard script for estimating the Klein model
> (klein.inp from the Practice file... submenu) I get a popup window
> saying
>
> 'I' shadows a function with the same name
>
> But I is a variable and not a function. So I don't see what the
> problem is.
We discussed that point not so long ago on this list. The potential
problem has to do with lags, as in "I(-1)" : such constructions are
hard to distinguish from function calls.
In this case actually could sometimes be impossible I guess: Is "I(2)" a
two-by-two identity matrix or the second lead of the series "I"? Who
knows...
Probably one wouldn't choose such a one-letter name 'I' for a built-in
function anymore because of these shadowing issues, but this function is
so widespread it cannot be easily replaced.
However, the error message leaves the user puzzled I guess. First of
all, what is 'I' supposed to refer to? If you don't know the script or
the dataset well it's not clear. Perhaps better to say "The dataset
series 'I' shadows...".
Secondly, what is the consequence? There is no hint as to what is going
to happen or why it matters. So perhaps add a second sentence in the
warning like "Console or script commands might be wrongly interpreted."
Agreed, the behavior you describe is not nice and it should be
improved.
Also agreed.
cheers,
sven