2003 article about Gretl is outdfated: now Gretl is much, much better than described there.
Also, Carlos Henrique Coêlho de Andrade from Divisão de Macroeconomia do Banco do Brasil
is a Gretl translator.
I have seen several articles "Econometric modelling in National Bank of X country"
It would be nice if Carlos would be invited to make Gretl working paper on this topic
Gretl is buggy:
Most of Gretl bugs are at developer level and typically they have a way around.
Gretl once had have a common bug with e-Views in using information criteria
for lag selection. In Gretl it is corrected, e-Views???
Once I have encountered a really serious bug in e-Views: for some dataset
they gave estimates of an ARMA model corresponding to non-invertible
MA-part version.
I use Gretl in some classes: for some things I wrote R-interfaces (unwanted by Riccardo
as official packages)
I use R for some classes: for some topics I simply give my Gretl materials.
I tell students: if you are working in a small firm it's stupid to by, say SAS.
If you working in an institution with a given soft it would be much easier
to learn it after having acquainted with two different programs.
It is commercially bad to integrate, say Stata and e-Views.
But integration Gretl with, say R and python is quite natural.
Oleh