Re: [Gretl-devel] grabbing Alfred

Alternative approach: in the attached tarball you'll find 1) an unmodified file from Alfred (if you choose the "Observations by Real-Time Period" format, like Sven did) 2) a hansl script (alfred_test.inp), which works on the file above, and performs all the required preprocessing without having to resort to Python. This file will give an error at the "join" command, which is the same error that Sven got originally; note that, by using the --tkey and --time options, the outer key field is ok "as-it-comes". 3) a diff file which, if applied to CVS, will make the source of the error evident: in practice, the string->codes transformation of the string variable used for filtering is not done properly. I'd like to point out that (3) is a bug we'll have to fix anyway, whatever solution we agree upon in order to code missing values in the outer file. Unfortunately, I won't be able to work on this today or tomorrow.