2:53 a.m.
Hi all,
Since it seems that the list is a bit quit today, I thought I should send something for
Allin to work on. :)
Fred
Over time, I have noticed a few items that I think might be bugs of gretl.
1. window list issue.
2. time series data with very long names.
3. lagged variable might not show in the main window.
4. can’t see command log whenever I’m working on .gretl files
5. "Show main statistics" vs. "Show full statistics" option dialog
often were hidden behind the icon window when one ask for the summary statistics.
6. If a model was saved with a name that contains “&”, in the icon view, once the
session is saved and then if one tries to open that session, the model with name that
contains “&” con’t be recovered. Either one can’t save a model with names that
contains “&”, or gretl should be able to recover that model, it seems.
Here are the details:
1. After a script is opened, or a new one is created, if I save it with a new name, the
list of windows that are open when I click on the top right corner of gretl still shows
the old name, which can cause confusion.
———————————————————————————————————
2. With time series data, if the names of two variables that are very long, close to the
31 character limit, and they differ only at the last one or two characters, then a
regression of the two with lagged variables will clash with gretl auto created shortened
abbreviations.
For example, if I run this script first on a new dataset with only the following two
variables,
# gretl script 1
ols o123456789012345678901234567890 0 o1234567890123456789012345678(-1 to \
-4) o123456789012345678901234567890(-1 to -4) —robust
I get this gretl error message:
gretl version 2019b-git
Current session: 2019-04-12 09:26
? ols o123456789012345678901234567890 0 o1234567890123456789012345678(-1 to \
-4) o123456789012345678901234567890(-1 to -4) --robust
variable 3 duplicated in the command list.
Error executing script: halting
ols o123456789012345678901234567890 0
o1234567890123456789012345678(-1 to -4) o123456789012345678901234567890(-1 to -4) —robust
upon inspecting the dataset, I get:
? list
Listing 7 variables:
0) const
1) o123456789012345678901234567890
2) o1234567890123456789012345678
3) o12345678901234567890123456_1
4) o12345678901234567890123456_2
5) o12345678901234567890123456_3
6) o12345678901234567890123456_4
? labels 3 4 5 6
Listing labels for variables:
o12345678901234567890123456_1: = o1234567890123456789012345678(t - 1)
o12345678901234567890123456_2: = o1234567890123456789012345678(t - 2)
o12345678901234567890123456_3: = o1234567890123456789012345678(t - 3)
o12345678901234567890123456_4: = o1234567890123456789012345678(t - 4)
I then renamed it down to a shorter names
rename 2 o12345678901234567890123456
rename 1 o1234567890123456789012345678
Listing 3 variables:
0) const
1) o1234567890123456789012345678
2) o12345678901234567890123456
? ols 1 0 o1234567890123456789012345678(-1 to -4) \
o12345678901234567890123456(-1 to -4)
variable 3 duplicated in the command list.
Error executing script: halting
ols 1 0 o1234567890123456789012345678(-1 to -4)
o12345678901234567890123456(-1 to -4)
gretl console: type 'help' for a list of commands
? list
Listing 7 variables:
0) const
1) o1234567890123456789012345678
2) o12345678901234567890123456
3) o12345678901234567890123456_1
4) o12345678901234567890123456_2
5) o12345678901234567890123456_3
6) o12345678901234567890123456_4
? labels 3 4 5 6
Listing labels for variables:
o12345678901234567890123456_1: = o12345678901234567890123456(t - 1)
o12345678901234567890123456_2: = o12345678901234567890123456(t - 2)
o12345678901234567890123456_3: = o12345678901234567890123456(t - 3)
o12345678901234567890123456_4: = o12345678901234567890123456(t - 4)
So it seems that the issue is the inconsistent naming of the lagged variables.
———————————————————————————————————
3. Sometimes the lagged variable don’t show up in the main window, so that I can’t delete
those lagged ones, other then using the list command and then invoke the delete varlist
command.
———————————————————————————————————
4. When I’m working on a .gretl file, if I choose to see the command log, an error would
show up.
Sometimes it says “Data error”, while other times, it says:
"Error opening directory '…’: No such file or directory"
Error opening directory
'/Volumes/FredMacHD/Users/fred/Documents/UIBE/Resource-Econometric/Dataset/gretl/functions':
No such file or directory
Error opening directory '/Volumes/FredMacHD/Users/fred/gretl/functions': No such
file or directory
Error opening directory
'/Volumes/FredMacHD/Users/fred/Documents/UIBE/Resource-Econometric/Dataset/gretl/functions':
No such file or directory
Error opening directory '/Volumes/FredMacHD/Users/fred/gretl/functions': No such
file or directory
Error opening directory …
———————————————————————————————————
5. Once I open the icon view, if I double click on the summary icon, often I see nothing
is happening. Once I move the icon view to somewhere else, there is the dialog box asking
for options to summarize. It was hidden behind the icon view window the whole time.
Fred
9:28 a.m.
Am 01.05.2019 um 04:53 schrieb Fred Engst:
Here are the details:
1. After a script is opened, or a new one is
created, if I save it with a new name, the list of windows that are
open when I click on the top right corner of gretl still shows the
old name, which can cause confusion.
Hm, I only see "script editor" there anyway, but perhaps that has to do
with tabs activated for me. Are you using the script editor in a tabbed way?
———————————————————————————————————
2. With time series data, if the names of two variables that are very
long, close to the 31 character limit, and they differ only at the
last one or two characters, then a regression of the two with lagged
variables will clash with gretl auto created shortened
abbreviations.
Certainly a limitation, not sure if it counts as a bug. As the person
who called for the extension of variable names years ago (from a much
shorter limit) I see what you mean, but I guess it's OK to ask the user
to stay below 28 or 25 characters or so.
OTOH when series are auto-imported (dbnomics!) and the long names come
from somewhere else, the issue may be more problematic.
I then renamed it down to a shorter names rename 2
o12345678901234567890123456 rename 1 o1234567890123456789012345678
Listing 3 variables: 0) const 1) o1234567890123456789012345678 2)
o12345678901234567890123456
? ols 1 0 o1234567890123456789012345678(-1 to -4) \
o12345678901234567890123456(-1 to -4) variable 3 duplicated in the
command list.
Error executing script: halting
Those names are still very long, not sure what your point is here, given
the limitation above.
———————————————————————————————————
3. Sometimes the lagged variable don’t show up in the main window, so
that I can’t delete those lagged ones, other then using the list
command and then invoke the delete varlist command.
My guess is that they're "hidden" below their parent variable, watch out
for a little arrow or triangle symbol in the variable list. When you
click on it the lags are shown.
———————————————————————————————————
4. When I’m working on a .gretl file, if I choose to see the command
log, an error would show up.
Confirmed (I think). My guess is that it has to do with gretl's
"philosophy": Either use the command log to record your actions, or use
a session-file based approach (.gretl), but not both. But first this is
a bit of speculation, and secondly even then it should be caught (or
disabled).
———————————————————————————————————
5. Once I open the icon view, if I double click on the summary icon,
often I see nothing is happening. Once I move the icon view to
somewhere else, there is the dialog box asking for options to
summarize. It was hidden behind the icon view window the whole time.
Cannot confirm, but I believe you ;-)
Here we need more info about your setup, OS, version etc.
Thanks for the reports
sven
11:06 p.m.
On Wed, 1 May 2019, Fred Engst wrote:
Since it seems that the list is a bit quit today, I thought I
should send something for Allin to work on. :)
Over time, I have noticed a few items that I think might be bugs
of gretl.
Ok, here's the scorecard so far.
1. window list issue.
Still under consideration.
2. time series data with very long names.
That is, insane identifiers! Under consideration.
3. lagged variable might not show in the main window.
Not examined yet. Probably linked with point 2?
4. can’t see command log whenever I’m working on .gretl files
Point taken. Should now be fixed, I think.
5. "Show main statistics" vs. "Show full
statistics" option dialog
often were hidden behind the icon window when one ask for the
summary statistics.
Should be fixed. The "z-order" (in front/behind) of GTK windows
"just works" on Linux but doesn't always work right on Mac or
Windows. I've made an attempt to force this in the right direction
for the option dialog in question.
6. If a model was saved with a name that contains “&”, in the
icon
view, once the session is saved and then if one tries to open that
session, the model with name that contains “&” con’t be recovered.
Either one can’t save a model with names that contains “&”, or
gretl should be able to recover that model, it seems.
Definitely fixed. We were "escaping" XML-special characters such as
"&" (&) in some contexts, but not all. It's now done for model
names. (But this will work only for newly saved session files; for
existing *.gretl files with ampersand in model names the damage is
done, I'm afraid.)
Allin
4:21 p.m.
On Wed, 1 May 2019, Fred Engst wrote:
After a script is opened, or a new one is created, if I save it with
a new name, the list of windows that are open when I click on the
top right corner of gretl still shows the old name, which can cause
confusion.
I see what you mean, and that's now fixed in git and snapshots.
(The problem was specific to the single-script editor window. If you
choose to use a tabbed script editor window, the window-list entry
just says "Script editor". But with the single window the name of the
script being edited is displayed in the window list, and this was not
updating in response to "Save as...".)
Allin
4:37 p.m.
On Wed, 1 May 2019, Sven Schreiber wrote:
Am 01.05.2019 um 04:53 schrieb Fred Engst:
> 2. With time series data, if the names of two variables that are very
> long, close to the 31 character limit, and they differ only at the
> last one or two characters, then a regression of the two with lagged
> variables will clash with gretl auto created shortened
> abbreviations.
Certainly a limitation, not sure if it counts as a bug. As the person
who called for the extension of variable names years ago (from a much
shorter limit) I see what you mean, but I guess it's OK to ask the user
to stay below 28 or 25 characters or so.
Yes, I think we should place a limit of (say) 22 or 24 bytes on
user-specified series names. This would leave room for the prefixes
and suffixes that get added automatically in generating logs, squares,
lags and so on.
If we do that I guess we could still accept names of up to 31 bytes in
third-party data files, but should probably issue a warning.
Allin
1824
days inactive
1825
days old
4 comments
3 participants
participants (3)
-
Allin Cottrell -
Fred Engst -
Sven Schreiber