there seems to be a change in Gretl 2017b that leads to failure of the comprehensive estimation and test script which is available at: http://www.boeckler.de/pdf/p_imk_study_54_2017_replikation.zip This replication script runs fine on Gretl 2017a but fails with an out of memory error in 2017b and the latest snapshot (or, as Artur tells me, with recent git on Linux). Could this be a new gretl bug? Is there a workaround for 2017b?

On Tue, 13 Jun 2017, Sven Schreiber wrote: > there seems to be a change in Gretl 2017b that leads to failure of the > comprehensive estimation and test script which is available at: > > http://www.boeckler.de/pdf/p_imk_study_54_2017_replikation.zip > > This replication script runs fine on Gretl 2017a but fails with an out > of memory error in 2017b and the latest snapshot (or, as Artur tells > me, with recent git on Linux). > > Could this be a new gretl bug? Is there a workaround for 2017b? Yes, a bug. It's now fixed in git and snapshots. There is a workaround for the 2017b release. The trouble arose from repeated assignment of a constant string to a bundle in a loop, as in loop N bundle b = null b.str = "foo" endloop Works the first time, then fails. But if you substitute b.str = sprintf("foo") it works OK. Just to be clear, this hack is not required with current git/snapshot. Allin