[Gretl-users] Re: Error executing the convolve.inp script file
On Sat, 25 Jul 2026, Sven Schreiber wrote:
Am 25.07.2026 um 10:49 schrieb Artur Bala:
> Thank you Allin,
>
> I tested the new script and it works fine.
>
> Thinking out loud, for outdated commands - like cmult() - could
> the « help section » also mention that the command is no longer
> used? When a script fails users’ first reaction is to type « help
> xyz » to check for coding errors (as I did in this case but
> everything looked ok).
>
Well, at least on current gretl versions, when I type "help cmult",
the function reference is opened, and there in the cmult entry it
says "This is a legacy function, predating gretl's native support
for complex matrices."
There's also the changelog with backwards-incompatible changes, and
sometimes there are typically deprecation notices. However, "cmult"
hasn't been officially deprecated it seems, just superseded. Maybe
it's time to do that.
Plus, it's still not fully clear to me whether the error in the
outdated script was legit (although perhaps not very clear) or
really buggy. I mean if that example script used to work and cmult()
is still being dragged around, then I guess the script should
ideally still work?
cmult() still works, given old-style (pre-2020) complex matrices as
input. The trouble with the outdated convolve script was that fft()
now produces a new-style (proper) complex result. I'd say that cmult()
and cdiv() are both pretty much useless at this point. You could cook
up old-style complex matrices and feed them to these functions as
arguments, but why would anyone want to do that?
Allin