Re: [Gretl-devel] Built-in spreadsheet improvement

On Tue, 16 Feb 2010, [ISO-8859-1] Patricio Cuar�n wrote: I'll just address the first suggestion at this point. I don't mean to be defensive, but the statement above is not accurate. If gretl can determine, from the data file to be appended, a set of observation numbers/dates that makes sense in relation to the original data, then the append goes through fine even though the observation ranges are not the same. Furthermore, gretl bends over backwards to to try parse sensible observation numbers/dates out of the file to be imported. (As stated in the manual, the observations column should be headed "obs" or "date" to help gretl.) Here are some examples that I just verified. First, we open the annual data file data3-6.gdt, which runs from 1959 to 1994. Then we try appending ("/File/Append data/text/CSV...") the file whose content is as follows: year,z1,z2 1970,1,2 1971,2,3 1972,3,4 No problem, it's appended correctly, with the observations for the two added variables placed properly with regard to date. Now we try the quarterly data file data9-7.gdt, which runs from 1975:1 to 1990:4. We try appending in turn plain text files with the following content: 1) obs,x1,x2 1990:3,20,30 1990:4,30,40 1991:1,40,50 2) obs x1 x2 1990Q3 20 30 1990Q4 30 40 1991Q1 40 50 3) obs x1 x2 1991.1 20 30 1991.2 30 40 1991.3 40 50 All of these are handled correctly, whether comma-separated or space separated, whether the year/quarter boundary is represented by ":", "Q" or ".", and whether the new data overlap the original observations and extend the dataset by by one observation (the first two cases), or "stick onto the end" of the original data and extend the dataet by the full number of new observations (case 3). Nothing in this regard has been changed since the 1.8.7 release, so I'm not drawing on brand-new CVS behavior. Allin Cottrell