One nasty thing you might be able to do in the interim is to shimmer the
object to the type you want by doing something like [set x [expr {double
$x}]] to make it double.
On Wed, Feb 27, 2019 at 9:31 AM Richard Hipp <d...@sqlite.org> wrote:
> On 2/27/19, Jean-Baptiste Gardette <jb.garde...@orens.fr> wrote:
> > have you any explanation why the typof() output of
> > "SELECT DerefDict_SQL('A'), typeof(DerefDict_SQL('A'))"
> > gives "text" in Sequence A and "real" in Sequence B ?
> >
>
> No.
>
> The code that determines the return type of a TCL UDF is here:
> https://www.sqlite.org/src/artifact/6b19e7562195aaf8?ln=993-1018
>
> As you can see, the interface is looking at the Tcl_Obj.typePtr value
> to try to guess an appropriate return type. This is considered bad
> form in TCL since TCL makes no guarantees about the value of typePtr
> for a Tcl_Obj. But the technique has worked well for nearly two
> decades.
>
> What if we were to add a new option to the "db function" method of the
> TCL interface that let you specify the return type of the
> TCL-implemented UDF? If the optional argument is omitted, then the
> current behavior of trying to guess the return type based on typePtr
> is used. But if the "-returntype TYPE" argument is present, it
> determines which return type to use regardless of the value in the
> typePtr. Would that help your situation?
>
> Note that my tests were run after compiling both TCL and SQLite from
> sources. There is no telling what non-standard changes may have been
> added by people who assembled your pre-compiled binaries.
>
>
>
> --
> D. Richard Hipp
> d...@sqlite.org
> _______________________________________________
> sqlite-users mailing list
> sqlite-users@mailinglists.sqlite.org
> http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
>
_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users
