Re: [Gretl-users] Testing the current CVS version
Hi,
some comments about the latest windows snapshot (July 6) in preparation
for 1.6. I haven't done any thorough testing, though, and some comments
are quite general, i.e. could have been made earlier as well:
1) gretl still uses c:\userdata in various places, even though it was
installed in c:\programme\gretl in my case. I guess it should use a
subdir of the install dir, or try to figure out the home dir of the user
2) The message "there are no new files on the server right now" is still
not localizable.
3) In preferences/general, the choice between cholesky and QR could
maybe use a short comment what this is about (I guess it must relate to
how to solve (X'X)*beta = X'y ?)
4) Just a thought: maybe use 2sls instead of tsls, in analogy to 3sls,
and both "two" and "three" start with t.
5) I like the new menu structure; however, iirc there used to be one
menu less. Now the session menu is quite short, maybe make that a
submenu of file? (I haven't thought this through, but staying at 7 menus
would be better imho.)
6) Function packages: I just ran the Hegy example on my cross-section
dataset. So maybe it would be a good idea to have function authors
specify the type of datasets for which their function is applicable, and
otherwise refuse to run. (Of course, users can be expected to not do
something so stupid, but I guess it would potentially also catch other
data-management errors.) And for function uploading I guess you made
sure that overwriting existing functions is impossible?
7) Whether or not variables are (set to) discrete should be displayed in
the main window (by means of a little symbol, or a letter). And for
discrete variables it would be nice to replace the histogram by a bar
diagram where the bars are visually separated.
8) The message string 'Data set is not recognized as a panel.\n
Please use \"Sample/Set frequency, startobs\".' still needs to be changed.
9) In the Hurst exponent window there are some English strings that
apparently are not localizable yet: "coeff", "std. error",
"Intercept",
"Slope", "avg", "Size" (2x).
That's it for now,
cheers,
Sven