1:58 p.m.
A minor annoyance:
When making a new function package, the function package editor also
lists obsoleted function names.
That is, if I run the following script three times:
function example (void)
print "hello"
end function
I will get three entries named "example" in the function package editor,
where there should be only one (referring to the latest definition of
the function).
-Sven
(ps: empty lists in the function package gui seem to work well!)
2:01 p.m.
On Mon, 15 Jan 2007, Sven Schreiber wrote:
When making a new function package, the function package editor also
lists obsoleted function names.
That is, if I run the following script three times:
function example (void)
print "hello"
end function
I will get three entries named "example" in the function package
editor, where there should be only one (referring to the latest
definition of the function).
OK, thanks for spotting that, I'll take a look.
Allin.
7:36 p.m.
On Mon, 15 Jan 2007, Sven Schreiber wrote:
if I run the following script three times:
function example (void)
print "hello"
end function
I will get three entries named "example" in the function package
editor, where there should be only one (referring to the latest
definition of the function).
How exactly are you running the script? I tried a few ways but
each time only found one instance of "example" in the packager
window. (I also get the message, "Redefining function 'example'",
on each re-run.)
Allin
9:10 p.m.
Allin Cottrell schrieb:
On Mon, 15 Jan 2007, Sven Schreiber wrote:
How exactly are you running the script? I tried a few ways but each
time only found one instance of "example" in the packager window. (I
also get the message, "Redefining function 'example'", on each
re-run.)
Ok it took me a little while to investigate, although it had repeatedly
happened to me in "real life"... The reason is that a function error is
necessary to trigger it -- something quite frequent with my poor
programming skills :-) So my previous report wasn't quite correct,
because there the function was fine and would never cause an error.
To replicate:
1. start gretl
2. open an arbitrary dataset
3. run a script with the following function:
function example(void)
print hu
end function
(note the syntax error)
4. package the function
5. execute the newly created function package => error (on the command
line it says "exiting on expr() error")
6. try to make a new function package...
... and you should see two 'example' entries in the function list.
HTH,
Sven
9:39 p.m.
On Tue, 16 Jan 2007, Sven Schreiber wrote:
Ok it took me a little while to investigate, although it had
repeatedly
happened to me in "real life"... The reason is that a function error is
necessary to trigger it...
Ah! Thanks for the detailed report, now this should be fixable
quite easily, I hope.
Allin.
9:17 p.m.
Sorry, this one slipped my mind.
On Tue, 16 Jan 2007, Sven Schreiber wrote:
To replicate:
1. start gretl
2. open an arbitrary dataset
3. run a script with the following function:
function example(void)
print hu
end function
(note the syntax error)
Hmm, it's not a syntax error. It's a semantic error (no variable
'hu' in scope). Unfortunately it's going to be difficult for
gretl to tell, in general, when a given error in function
execution is an indication of plain brokenness in the function (as
here), or is an error only in the context of the parameter values
and/or other conditionality.
4. package the function
5. execute the newly created function package => error (on the command
line it says "exiting on expr() error")
6. try to make a new function package...
... and you should see two 'example' entries in the function list.
Yes, I do, and that's bad. I'll do some digging.
Allin.
9:50 p.m.
On Mon, 22 Jan 2007, Allin Cottrell wrote:
> function example(void)
> print hu
> end function
Running this via the GUI now produces
'hu' is not the name of a variable
rather than "unspecified error". And we no longer end up with a
duplicated "example" function.
Thanks, Sven, I think we're getting somewhere with the
user-defined functions now.
Allin.
9:24 a.m.
Allin Cottrell schrieb:
Thanks, Sven, I think we're getting somewhere with the user-defined
functions now.
Thanks for all the quick bug fixes! The function packaging was already
nice from the beginning, that's why I stuck to it, the bugs really
weren't that bad. I think gretl's intuitive scripting capabilities are
really a strong selling point. Eviews' language for example may have
more features, but otherwise it is a nightmare IMHO.
I have now finished three function packages calling python/numpy in the
background (what I call py4gretl) and estimating some VECM stuff:
1) py4gretl_vecm allows to differentiate between restricted and
unrestricted exogenous terms, for example to have a level shift in the
data but not a broken trend. It returns standard estimation results, and
does tests for weak exogeneity and some tests on \alpha_\perp
(Gonzalo-Granger factor exclusion tests).
2) py4gretl_vecdecomp returns new series, namely the components of the
Gonzalo-Granger decomposition and the Stock-Watson common-trends
representation (in the Proietti variant, for those who want to know).
3) py4gretl_vecrestrict allows to place general linear restrictions on
\alpha and \beta, following Boswijk and Doornik's (2004, section 4.4)
switching algorithm.
Of course they need more testing and are definitely beta-quality. In
order to avoid confusion they officially require gretl 1.6.1, so I'm not
sure it makes sense to publish them before 1.6.1 is released.
There's one feature request I would have for 1.6.2 though: a matrix
editor accessible from the function package call dialog; that is,
whenever the user must choose a matrix as a function parameter, give her
the possibility to define/edit the matrix on the fly, similar to the
named list editing capabilities.
Cheers,
Sven
9:46 a.m.
On Tue, January 23, 2007 10:24, Sven Schreiber wrote:
I have now finished three function packages calling python/numpy in the
background (what I call py4gretl) and estimating some VECM stuff:
Yum Yum! Mummy, can I have it? Pleeease!!
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Facoltà di Economia "G. Fuà"
Ancona
5:20 p.m.
Riccardo (Jack) Lucchetti schrieb:
On Tue, January 23, 2007 10:24, Sven Schreiber wrote:
> I have now finished three function packages calling python/numpy in the
> background (what I call py4gretl) and estimating some VECM stuff:
Yum Yum! Mummy, can I have it? Pleeease!!
I would have sent it today if gretl hadn't eaten them (see other
post)... Just kidding, only the help texts are lost (and they weren't
too long), of course I have the script sources, just have to repackage.
I guess tomorrow.
-sven
5:30 p.m.
Sven Schreiber schrieb:
Riccardo (Jack) Lucchetti schrieb:
> On Tue, January 23, 2007 10:24, Sven Schreiber wrote:
>> I have now finished three function packages calling python/numpy in the
>> background (what I call py4gretl) and estimating some VECM stuff:
> Yum Yum! Mummy, can I have it? Pleeease!!
>
I would have sent it today if gretl hadn't eaten them (see other
post)... Just kidding, only the help texts are lost (and they weren't
too long), of course I have the script sources, just have to repackage.
I guess tomorrow.
Ok here are the three packages I have. You probably have to adjust the
p4g_call() helper functions in all packages to call the python version
that makes sense on your system (it's now set to python2.5 which is what
I currently need).
I hope the attachment comes through alright...
enjoy,
sven
6582
days inactive
6592
days old
10 comments
3 participants
participants (3)
-
Allin Cottrell -
Riccardo (Jack) Lucchetti -
Sven Schreiber