12:33 p.m.
Dear professors,
I'm using gretl2019a_git on windows 7 (64bit).
I usually save sessions with models to keep track of all the regressions
made.
I have 1 session saved with 10 OLS regressions and everything is ok.
I also have another session with 10 Fixed Effects regressions, which, when
opening such session, gives me the following error:
"10 session object(s) could not be rebuilt"
I've also tried to save OLS and FE regressions in the same session and it
rebuilds OLS' but not FE's.
The same problem occurred with gretl2018d.
Hope you can help me.
Thanks,
Marcelo
1:31 p.m.
Am 23.01.2019 um 13:33 schrieb Marcelo Duarte:
I'm using gretl2019a_git on windows 7 (64bit).
Yes, current snapshot here on
Win7 as well.
I've also tried to save OLS and FE regressions in the same session and
it rebuilds OLS' but not FE's.
I can reproduce the problem with the example abdata.gdt and a simple
bivariate FE model.
Apart from waiting for a bugfix I can only advise a workaround: Restart
gretl and use the command log that gretl records automatically and save
that as a script. (In the Tools menu.) There you will see the data file
you loaded ('open') and you will see your FE estimation command
('panel'). When you re-run that script you should get your results back.
Thanks for the report,
Sven
1:37 p.m.
On Wed, 23 Jan 2019, Sven Schreiber wrote:
Am 23.01.2019 um 13:33 schrieb Marcelo Duarte:
>
> I'm using gretl2019a_git on windows 7 (64bit).
Yes, current snapshot here on Win7 as well.
>
> I've also tried to save OLS and FE regressions in the same session and it
> rebuilds OLS' but not FE's.
>
I can reproduce the problem with the example abdata.gdt and a simple
bivariate FE model.
Hmm, I can't replicate the problem on Linux. I'll try on Windows asap.
Allin
1:57 p.m.
On Wed, 23 Jan 2019, Marcelo Duarte wrote:
Dear professors,
I'm using gretl2019a_git on windows 7 (64bit).
I usually save sessions with models to keep track of all the regressions
made.
I have 1 session saved with 10 OLS regressions and everything is ok.
I also have another session with 10 Fixed Effects regressions, which, when
opening such session, gives me the following error:
"10 session object(s) could not be rebuilt"
Problem found: it's due to the weird way that the Windows C library
writes nans (Not a Number). A fix will be forthcoming shortly.
Allin
8:45 p.m.
On Wed, 23 Jan 2019, Sven Schreiber wrote:
Am 23.01.2019 um 13:33 schrieb Marcelo Duarte:
>
> I'm using gretl2019a_git on windows 7 (64bit).
Yes, current snapshot here on Win7 as well.
>
> I've also tried to save OLS and FE regressions in the same session and it
> rebuilds OLS' but not FE's.
>
I can reproduce the problem with the example abdata.gdt and a simple
bivariate FE model.
Apart from waiting for a bugfix I can only advise a workaround: Restart gretl
and use the command log that gretl records automatically and save that as a
script. (In the Tools menu.) There you will see the data file you loaded
('open') and you will see your FE estimation command ('panel'). When you
re-run that script you should get your results back.
This issue is fixed in the 2019a release. In addition there's a fair
chance that the new release may successfully read saved models in
session files created with gretl 2018d or 2019a-git snapshots (I mean,
models that were not reconstructed properly before.)
Allin
3:03 p.m.
Am 24.01.2019 um 21:45 schrieb Allin Cottrell:
On Wed, 23 Jan 2019, Sven Schreiber wrote:
> I can reproduce the problem with the example abdata.gdt and a simple
> bivariate FE model.
>
This issue is fixed in the 2019a release.
Confirmed!
In addition there's a fair chance that the new release may
successfully read saved models in session files created with gretl
2018d or 2019a-git snapshots (I mean, models that were not
reconstructed properly before.)
Couldn't test that...
thanks,
sven
1917
days inactive
1919
days old
5 comments
3 participants
participants (3)
-
Allin Cottrell -
Marcelo Duarte -
Sven Schreiber