On Mon, 10 Sep 2007, andreas.rosenblad(a)ltv.se wrote: Unfortunately, CVS is not in my opinion in a releasable state yet. The main addition that needs ironing out is restricted VECM models, which is also undocumented in the manual (I'll try and find some time next week to work on this). What we _could_ do is ship a modified windows installer which contains the up-to-date windows dll of fftw, for those people who experience problems. However, it would be nice to nail this down properly and figure out exactly which windows versions exhibit the crash with fftw compiled in-house. Riccardo (Jack) Lucchetti Dipartimento di Economia Università Politecnica delle Marche r.lucchetti(a)univpm.it http://www.econ.univpm.it/lucchetti

On Mon, 10 Sep 2007, Sven Schreiber wrote: I tend to agree with Sven on this. It's true, we're not finished with the work on VECM restrictions, but I don't think we've broken anything in that area since 1.6.5. And we have fixed quite a few bugs across the board (see below). The thing I've been working on most recently is a re-vamping of various distribution-related items under the Tools menu (p-values, critical values, hypothesis tests, distribution graphs). Also moving the GUI random-variable generator over to the same interface as those other items. I've done some testing on this, but it could do with some more. I've also updated gretl.pot file in CVS. Cri has updated the Italian translation but the other still need doing. I'm appending the current ChangeLog. Allin. version 1.6.6, in progress - Add heckit estimator - Fix for loop on empty list - Fix for remaining issues with apostrophes in graph titles - Fix for lists as arguments for user-defined functions - Fix for saved sessions after using Forecast menu items - Fix for long variable names in Model Graph menus - Improve formatting for nonparametric difference test - Add "ceil" function in genr - Add "--p-values" option to show p-values instead of slopes for logit, probit models - Add options to GUI for xtab command (cross-tabulation); also add Fisher's Exact Test for 2 x 2 tables - Fix for graphing the time trend, and for graphs displaying more than 8 series - Fix for length of correlogram on weekly data - Fix for null matrix arguments to user functions in GUI - Fix for floating point values in "set" when using decimal comma - Fix for opening cross-sectional gretl data sets in R version 2.5.0 or higher - Fixes for autoregressive genr formula: speed-up, and fix bug when such formulae are used in an nls/mle block - Double maximum length to gretl command line to 8192 bytes - Gamma p-values: switch to shape/scale parameterization - Spearman's rho: fix for case with ties - Allow for syntax change re. missing values in gnuplot 4.3 - "corr" command: add --spearman and --kendall options - Rank correlation in GUI: add Kendall's tau option - Fix a few issues with extreme values in cdf, pvalue code - add gamma() function in genr to generate gamma random variable - Modify Locke's test to use Kendall's tau - Add ginv() (Moore-Penrose inverse) matrix function - Add multivariate Portmanteau test for autocorrelation in VARs (Ljung-Box) - Allow concatenation of multiple short-form options, as in "-rdq" - Modify retrieval of info from VARs: $coeff gets the matrix of coefficients, $compan gets the companion matrix - gretlcli: Ctrl-X exits without prompting for save - Databases: allow selection of more than one series for display, graphing, importing - Databases: allowing deletion of series from native gretl databases (if the user has write access) - New command, "dataset": subsumes the old "addobs" and "transpos", while adding new options - fftw DLL for Windows: fix a problem experienced on some Windows systems

On Mon, 17 Sep 2007, Sven Schreiber wrote: Thanks, that's now fixed in CVS. Allin.