[Gretl-devel] Re: the new VAR forecasting bug
On Fri, Oct 16, 2026 at 11:10 AM Sven Schreiber
<sven.schreiber(a)fu-berlin.de> wrote:
Hi all,
some may have noticed the following report about a bug which Allin has
already fixed in the meantime; https://sourceforge.net/p/gretl/bugs/337/
I have quickly tested this with gretl 2026b, and it seems to work there,
so that would be a regression in the recent 2026c release. I wonder if a
crash with VAR forecasting (with exog. terms) is bad enough to warrant
an emergency fix?
I've re-tested that issue and in fact inclusion of exogenous variables
is not required to trigger a crash. It would happen with any attempt
to forecast from a window holding VAR results in the GUI program. I
would not be opposed to offering a quick release of 2026d.
Allin