5:43 p.m.
Hi everybody,
so right after the conference I am trying to get this function package
about permanent-transitory decompositions in shape, and I am stumbling
over this issue in the following example code:
[to avoid confusion, "hansl" seems to emerge as the new name of gretl's
scripting language, a pseudo-acronym for HANdy Scripting Language]
<hansl>
function void hello(void)
print "do we get here?"
dummy = 0
print "and here?"
funcerr "poor function, only there to fail"
print "and then after funcerr?"
end function
hello()
</hansl>
In this code, gretl apparently stumbles over the 'funcerr' line. I also
noticed that 'funcerr' does not appear in the command or function
reference lists. (But it is in the manual.) Can somebody tell me what is
wrong here? The error message only says syntax error.
thanks,
sven
5:04 a.m.
On Sat, 18 Jun 2011, Sven Schreiber wrote:
Hi everybody,
so right after the conference I am trying to get this function package
about permanent-transitory decompositions in shape, and I am stumbling
over this issue in the following example code:
[...]
If you apply this patch, funcerr ought to work; I haven't committed this
to CVS yet, cause I wasn't able to trace when and why the call to
get_funcerr_message() disappeared, so I'll leave it to Allin to provide
a proper fix.
Index: lib/src/gretl_func.c
===================================================================
RCS file: /cvsroot/gretl/gretl/lib/src/gretl_func.c,v
retrieving revision 1.368
diff -u -w -r1.368 gretl_func.c
--- lib/src/gretl_func.c 2 Jun 2011 10:00:50 -0000 1.368
+++ lib/src/gretl_func.c 19 Jun 2011 08:53:01 -0000
@@ -6668,6 +6668,7 @@
} else {
gretl_if_state_reset(indent0);
}
+ pprintf(prn, get_funcerr_message());
} else if (retline >= 0) {
/* we returned prior to the end of the function */
gretl_if_state_reset(indent0);
[to avoid confusion, "hansl" seems to emerge as the new
name of gretl's
scripting language, a pseudo-acronym for HANdy Scripting Language]
For people who weren't at the conference: Sven suggested to give a name to
our scripting language, to which everybody agreed. Lee came out with HANSL
(as in HANdy Scripting Language), which was an instant success; Lee's
proposal was further refined by Allin as "Hansl's A Neat Scripting
Language", which I find irresistibly geeky.
So, from now on, any time you see the HANSL language mentioned in this
list (or, I suspect, anywhere else), don't panic: it's just what we used
to call "gretl's scripting language".
I suggest that someone adds HANSL to http://99-bottles-of-beer.net/
Riccardo (Jack) Lucchetti
Dipartimento di Economia
Università Politecnica delle Marche
r.lucchetti(a)univpm.it
http://www.econ.univpm.it/lucchetti
9:16 a.m.
On 06/19/2011 11:04 AM, Riccardo (Jack) Lucchetti wrote:
On Sat, 18 Jun 2011, Sven Schreiber wrote:
> Hi everybody,
>
> so right after the conference I am trying to get this function package
> about permanent-transitory decompositions in shape, and I am stumbling
> over this issue in the following example code:
[...]
If you apply this patch, funcerr ought to work; I haven't committed this
to CVS yet, cause I wasn't able to trace when and why the call to
get_funcerr_message() disappeared, so I'll leave it to Allin to provide
a proper fix.
Well it's also in 1.9.5 (I previously worked on current cvs). From the
changelog it seems it may have happened when addons and/or errmsg() were
introduced, between 1.9.3 and 1.9.4.
Anyway, Allin said he has a collection of test scripts he runs before
each release. What about making them public and letting other people
(e.g. from this list) add more test scripts? One of the goals would be
to cover all commands and functions, which would catch problems like the
present one. I guess the sourceforge site could be used for that, for
example introducing another tracker instance.
cheers,
sven
5:05 p.m.
Em 20 de junho 2011 Sven escreveu:
(...)
Anyway, Allin said he has a collection of test scripts he runs before
each release. What about making them public and letting other people
(e.g. from this list) add more test scripts? One of the goals would be
to cover all commands and functions, which would catch problems like the
present one. I guess the sourceforge site could be used for that, for
example introducing another tracker instance.
Dear Sven, I would like to have a collection on test scripts in order to
help Riccardo
and Allin to find potential bugs.
Best regards,
*Henrique C. de Andrade*
Doutorando em Economia Aplicada
Universidade Federal do Rio Grande do Sul
www.ufrgs.br/ppge
5:55 a.m.
On Mon, Jun 20, 2011 at 2:16 PM, Sven Schreiber <svetosch(a)gmx.net> wrote:
On 06/19/2011 11:04 AM, Riccardo (Jack) Lucchetti wrote:
> On Sat, 18 Jun 2011, Sven Schreiber wrote:
>
>> so right after the conference I am trying to get this function package
>> about permanent-transitory decompositions in shape, and I am stumbling
>> over this issue in the following example code:
>
> [...]
>
> If you apply this patch, funcerr ought to work; I haven't committed this
> to CVS yet, cause I wasn't able to trace when and why the call to
> get_funcerr_message() disappeared, so I'll leave it to Allin to provide
> a proper fix.
Well it's also in 1.9.5 (I previously worked on current cvs). From the
changelog it seems it may have happened when addons and/or errmsg() were
introduced, between 1.9.3 and 1.9.4.
Yes, Sven's test script works on 1.9.3 but not on 1.9.4. Thanks to Jack for
the fix, but I'd like to get to the bottom of this problem. I'll try
to put something
into CVS in the next day or two.
Anyway, Allin said he has a collection of test scripts he runs
before
each release. What about making them public and letting other people
(e.g. from this list) add more test scripts? One of the goals would be
to cover all commands and functions, which would catch problems like the
present one.
OK, I'll find a way to publish the test scripts. The trouble in this case is,
as it stands, the collection of scripts does not contain much in the way
of testing of error conditions. But clearly there should be at least one
script that exercises funcerr.
Following Sven's later suggestion, I would be in favour of releasing 1.9.6
once the funcerr problem is fixed.
Allin
9:14 a.m.
On Fri, Jun 24, 2011 at 10:55 AM, Cottrell, Allin <cottrell(a)wfu.edu> wrote:
Sven's test script works on 1.9.3 but not on 1.9.4. Thanks to Jack for
the fix, but I'd like to get to the bottom of this problem. I'll try
to put something into CVS in the next day or two.
I've now committed some changes that are intended both to fix and
to simplify the "funcerr" mechanism. I'm uploading new snapshots to
match.
Naturally, testing is needed!
Allin
7:28 p.m.
On 06/24/2011 03:14 PM, Cottrell, Allin wrote:
I've now committed some changes that are intended both to fix and
to simplify the "funcerr" mechanism. I'm uploading new snapshots to
match.
On current cvs the 'funcerr' issues seem to have disappeared! But I
still cannot post my prototype function package yet because I got
another error now (with some arbitrary data and parameter constellation).
good night,
sven
2:02 p.m.
Am 19.06.2011 11:04, schrieb Riccardo (Jack) Lucchetti:
On Sat, 18 Jun 2011, Sven Schreiber wrote:
> Hi everybody,
>
> so right after the conference I am trying to get this function package
> about permanent-transitory decompositions in shape, and I am stumbling
> over this issue in the following example code:
[...]
If you apply this patch, funcerr ought to work; I haven't committed this
to CVS yet, cause I wasn't able to trace when and why the call to
get_funcerr_message() disappeared, so I'll leave it to Allin to provide
a proper fix.
At the risk of stating the obvious, I think it's a pretty serious bug,
since any function package with 'funcerr' will not work on recent gretl
versions. Could 1.9.6 be released soon?
cheers,
sven
4895
days inactive
4901
days old
7 comments
4 participants
participants (4)
-
Cottrell, Allin -
Henrique Andrade -
Riccardo (Jack) Lucchetti -
Sven Schreiber