is the "no_const" specification for the ui-maker apparatus for a function package GUI really working as intended? I seem to have a case where I have a ui-maker function that requests no_const for a list argument X, but I can execute the function with X = const alright. BTW, X is set to have a "null" default, if that matters. Or does no_const really mean that any constant term is removed quietly? I don't have time to give a minimal example right now, but I could do so on Friday or so, if necessary.

Am 24.07.2024 um 22:45 schrieb Cottrell, Allin: > On Wed, Jul 24, 2024 at 4:22 PM Sven Schreiber > <sven.schreiber(a)fu-berlin.de&gt; wrote: >> I don't have time to give a minimal example right now, but I could do so >> on Friday or so, if necessary. > Please do. This may be redundant, but of course the "no_const" > specification only applies to GUI usage. Will do, and yes, GUI context.

On Fri, Jul 26, 2024 at 10:15 AM Sven Schreiber <sven.schreiber(a)fu-berlin.de&gt; wrote: > - The package specifies "no-data-ok" and this also shows up in the info. However, when executing the package GUI, a warning dialog with "this requires a dataset" appears. I think that's because your package depends on SVAR, which requires time-series data.

> - The GUI_SVAR_setID_uimaker function is declared as the ui-maker, and requests no_const both for the lists X and Y, as well as no_singleton for Y. However, in the GUI I can execute the package under all those cases that are supposed to be prevented. Your package doesn't seem to specify a menu attachment: that's required for the ui-maker to kick in.

Am 26.07.2024 um 19:30 schrieb Cottrell, Allin: > On Fri, Jul 26, 2024 at 10:15 AM Sven Schreiber > <sven.schreiber(a)fu-berlin.de&gt; wrote:

>> - The package specifies "no-data-ok" and this also shows up in the info. >> However, when executing the package GUI, a warning dialog with "this >> requires a dataset" appears.

> I think that's because your package depends on SVAR, which requires > time-series data.

Ah, good point. Although then the follow-up question why is it just asking for an unspecified dataset, and not for time series?

>> - The GUI_SVAR_setID_uimaker function is declared as the ui-maker, and >> requests no_const both for the lists X and Y, as well as no_singleton for >> Y. However, in the GUI I can execute the package under all those cases >> that are supposed to be prevented.

Am 28.07.2024 um 02:52 schrieb Allin Cottrell: > Ah, your ui-maker specifications for list arguments are out-of-date. > The possibilities are: list_no_singleton, list_no_const and > list_exclude. IIRC it was you who requested the "list_" prefix for > these conditions ;-). See the current function-package doc. ... So that means we have to adjust the (few) packages that use the old style? For example, StrucTiSM does, and maybe some others, too.

On Tue, Jul 30, 2024 at 4:34 AM Sven Schreiber <sven.schreiber(a)fu-berlin.de&gt; wrote: > Am 28.07.2024 um 14:33 schrieb Sven Schreiber: >> So that means we have to adjust the (few) packages that use the old >> style? For example, StrucTiSM does, and maybe some others, too. > Is this no_singleton vs. list_no_singleton thing visible in the XML code > in the gfn file? Yes, you just have to look at the ui-maker function.

Am 30.07.2024 um 14:14 schrieb Cottrell, Allin: > On Tue, Jul 30, 2024 at 4:34 AM Sven Schreiber > <sven.schreiber(a)fu-berlin.de&gt; wrote: >> Am 28.07.2024 um 14:33 schrieb Sven Schreiber: >>> Am 28.07.2024 um 02:52 schrieb Allin Cottrell: >>>> Ah, your ui-maker specifications for list arguments are out-of-date. >>>> The possibilities are: list_no_singleton, list_no_const and >>>> list_exclude. IIRC it was you who requested the "list_" prefix for >>>> these conditions ;-). See the current function-package doc. >>> ... >>> So that means we have to adjust the (few) packages that use the old >>> style? For example, StrucTiSM does, and maybe some others, too. >> Is this no_singleton vs. list_no_singleton thing visible in the XML code >> in the gfn file? > Yes, you just have to look at the ui-maker function. OK, I checked all the function packages, and apart from the mentioned StrucTiSM the constructs without list_ are used in ParMA and in random_forest. However: I also read the package doc more thoroughly, and it says there: "These keywords can also be given in a shorter form, omitting the list_ prefix, ". That would then bring me back to the original question why it didn't seem to work in the example I gave? (I haven't tested yet whether it would work with the list_ prefix.)

On Tue, Jul 30, 2024 at 5:11 PM Cottrell, Allin <cottrell(a)wfu.edu&gt; wrote: > Good point, and good question. I'll take another look at this. It seems there was unfinished business in the relevant gretl source file, gui/fncall.c. That's now updated, and I think the behavior should be closer to what's claimed in the documentation, for both variants of the specifications pertaining to list arguments (that is, both with and without the "list_" prefix in "list_no_const", "list_no_singleton" and "list_exclude").

Am 05.08.2024 um 18:19 schrieb Sven Schreiber: > Am 04.08.2024 um 22:30 schrieb Cottrell, Allin: > > On Tue, Jul 30, 2024 at 5:11 PM Cottrell, Allin <cottrell(a)wfu.edu&gt; wrote: > > > > It seems there was unfinished business in the relevant gretl source > > file, gui/fncall.c. That's now updated, and I think the behavior > > should be closer to what's claimed in the documentation, for both > > variants of the specifications pertaining to list arguments (that is, > > both with and without the "list_" prefix in "list_no_const", > > "list_no_singleton" and "list_exclude"). > > Thanks, Allin, that sounds very good. (Although "closer to... the doc" is > also an interesting formulation.) So no need to update any existing > packages, if I understand correctly. > I've just tested the previous stripped down example gfn with the latest snapshot (GTK3). 1) The no_singleton check partly works, for the LHS (Y): - The drop-down menu in the GUI only shows existing lists, not single series. And creating a new list in that GUI context refuses to do a single-item list. However, pre-defining a singleton list and then selecting it does not yield any error message (down the road). Also, typing in the name of a single series does not raise an error (again, on clicking OK).

- For the other argument, which should honor the same restrictions, nothing seems to be prevented or restricted, or checked.

2) About the data requirement, the error message still says "need active dataset", not asking for time series. (Remember the package itself required nothing, but the provider SVAR wants time series.)