Dear all listers, greetings from Paris ! I have finally pinpointed my data problem. Here is a summary: As i wrote some time ago, I am still on a demand system analysis on large cross-sectional data from a french familly expenditures survey. Currently my data comes in a 10,240 x 581 matrix stored in standard gretl format (gzipped). I use scripts to perform estimations, elasticities computations followed by standard error estimation of elasticities via the delta method (using gretl matrix language). When i launch estimations the script perform all computations and creates a set of results variables for further analysis. Clearly everything runs smoothly. The trouble begins just there: if I save the dataset containing the results, in zipped format and I try to reopen it, gretl fails to open it. I tried several tricks, like saving a datafile that can be opened and that i save under a new name. I tried also to use console mode and store commands so that the data are not compressed. it works ok EXCEPT THAT when i use the store command on the full datafile + the created result variables, the file is created but i CANT OPEN IT. I finally decided to do just that: open the safe data file, execute my script, then export the identifiers, weights and ONLY the created variables. But when this is done in the GUI, the saved file is gzipped and when i try to open it fails. So it seems that there is a problem with the created variables and the gzip step. In effect gretl says: "Failed to parse data values at obs 391" lately it was at another observation too. I was advised to look at an XML reader, and i did just that after unzipping but it did not seem to be corrupted at the mentionned observation. I will try to export those variables in CSV or a non gretl format. any idea it seems there are problems with my created variables ? I am under win XP. thanx and nice WE chaps ! sincerely Franck

On Fri, 4 Jun 2010, Riccardo (Jack) Lucchetti wrote: Me too, please! Allin Cottrell