Re: [Gretl-users] bug-hunting time
On Tue, 6 Sep 2011, artur bala wrote:
This is not actually a bug but may be you'll find it useful to
point out.
Thanks, Artur. Yes, your comments are useful.
I'm currenty handling a large census database and:
1. I observed some limitations through the GUI:
a. adding new observations is limited to 100 00 obs;
b. creating a new data set is limited to 1 000 000 obs;
Otherwise, there's no limits to the "addobs" and
"nulldata"
commands in the CLI
I'm just trying to guard against the GUI becoming "frozen" due
to an excessive memory request. I guess I could raise the
limits but put up an "Are you sure?" dialog for values greater
than the current limits
2. Plotting the frequency distribution of a constant series X (GUI)
pops
up a very clear error message " 'X' is a constant". Instead, the
"freq"
command (CLI) ends up with 3 different error messages not so easy to
figure out.
You're right, that was a mess. It should be cleaned up now in
CVS.
In the same run, the boxplot command has the same behavior (GUI-CLI)
although the error message should perhaps, be more intuitive (like
frequency's "'X' is a constant" for example instead of
"C:\Program
Files\gretl\wgnuplot.exe"
"C:\Users\Artur\AppData\Roaming\gretl\gpttmp.i03604": exit code 1").
Same answer as above.
3. While executing commands in the CLI - specially loop iterations -
gretl seems to shut down but actually it doesn't.
Would be possible to show something like a progress bar (in fact, we
used to have a stuff like that in the script window) just to tell the
user that something is still going on?
We have something like that in the script window, namely a
"spinner" which should start going round when a time-consuming
script is being executed. It's not always possible for this to
move in real time because it's updated just whenever gretl
"comes up for air", but in my experiments it works reasonably
well.
Allin