6:18 p.m.
Dear all,
I noticed this thing.
(gretl 2022c on Windows)
<hansl>
strings s = null
s[""]
</hansl>
<output>
? strings s = null
? s[""]
Index value 969423488 is out of bounds
Error executing script: halting
s[""]
</output>
The number changes at every run. (maybe a memory address?)
The same problem occurs with arrays and bundles and changing the string.
As you can see it is not a serious problem but I think it's good to fix it.
Regards
Federico
6:30 p.m.
Am 02.11.22 um 19:18 schrieb Federico Fiorani:
Dear all,
I noticed this thing.
(gretl 2022c on Windows)
<hansl>
strings s = null
s[""]
</hansl>
<output>
? strings s = null
? s[""]
Index value 969423488 is out of bounds
Error executing script: halting
>s[""]
</output>
The number changes at every run. (maybe a memory address?)
The same problem occurs with arrays and bundles and changing the string.
As you can see it is not a serious problem but I think it's good to fix it.
Hi Fedrico,
thanks for the report.
Here on latest git version running this
<hansl>
clear
set verbose off
strings a = null
strings b = array(0)
print a b
# I think this is actually not supported
# and does not really look like a proper array access
catch a[""]
eval $error
catch b[""]
eval $error
# This is the way to go -- of course failing for an empty array
catch a[0]
eval $error
catch a[1]
eval $error
catch b[0]
eval $error
catch b[1]
eval $error
</hansl>
yields for each of those cases the error code 17.
Actually, I am surprised about a two things:
1. That this actually works: "strings a = null" I thought only "array(0)
works for initialisation.
2. That this command <a[""]> actually executes something. This, I would
argue, is not valid indexation of a string array where the index should
be a positive integer.
Artur
1:46 p.m.
On Wed, 2 Nov 2022, Federico Fiorani wrote:
Dear all,
I noticed this thing.
(gretl 2022c on Windows)
<hansl>
strings s = null
s[""]
</hansl>
<output>
? strings s = null
? s[""]
Index value 969423488 is out of bounds
Error executing script: halting
>s[""]
</output>
The number changes at every run. (maybe a memory address?)
No string (empty or otherwise) can be a valid index into an array (or
matrix, or string), so the example above should provoke a
non-conformable types error -- and now it does in git.
However, a string in square brackets can be a valid accessor for a
bundle element, and equivocation over that point explains why the
error wasn't being caught correctly before.
The same problem occurs with arrays and bundles
Hmm, I'm not seeing a problem with bundles. Even before today's fix I
was getting:
<hansl>
bundle b = null
b[""]
</hansl>
<output>
? bundle b = null
? b[""]
"": no such item
Terminated on error
</output>
which seems quite correct. What are you seeing with bundles?
Allin
2 p.m.
Il 03/11/2022 14:46, Allin Cottrell ha scritto:
On Wed, 2 Nov 2022, Federico Fiorani wrote:
> Dear all,
> I noticed this thing.
>
> (gretl 2022c on Windows)
> <hansl>
> strings s = null
> s[""]
> </hansl>
>
> <output>
> ? strings s = null
> ? s[""]
> Index value 969423488 is out of bounds
>
> Error executing script: halting
>> s[""]
> </output>
>
> The number changes at every run. (maybe a memory address?)
No string (empty or otherwise) can be a valid index into an array (or
matrix, or string), so the example above should provoke a
non-conformable types error -- and now it does in git.
Thanks
However, a string in square brackets can be a valid accessor for a
bundle element, and equivocation over that point explains why the
error wasn't being caught correctly before.
> The same problem occurs with arrays and bundles
Hmm, I'm not seeing a problem with bundles. Even before today's fix I
was getting:
<hansl>
bundle b = null
b[""]
</hansl>
<output>
? bundle b = null
? b[""]
"": no such item
Terminated on error
</output>
which seems quite correct. What are you seeing with bundles?
Sorry, I expressed
ambiguously. I meant "bundles" type, array of bundles.
(gretl 2022c)
<hansl>
bundles bun = null
bun[""]
</hansl>
<output>
? bundles bun = null
? bun[""]
Index value 2094668032 is out of bounds
Error executing script: halting
bun[""]
</output>
Federico
2:14 p.m.
Am 03.11.2022 um 15:00 schrieb Federico Fiorani:
Il 03/11/2022 14:46, Allin Cottrell ha scritto:
> On Wed, 2 Nov 2022, Federico Fiorani wrote:
>
>> The same problem occurs with arrays and bundles
> Hmm, I'm not seeing a problem with bundles. Even before today's fix I
> was getting:
>
Sorry, I expressed ambiguously. I meant "bundles" type, array of bundles.
(gretl 2022c)
<hansl>
bundles bun = null
bun[""]
</hansl>
Yes, that would be expected (given the previously existing bug), since
the "bundles" type is also an array. (As opposed to the "bundle"
type.)
cheers
sven
804
days inactive
805
days old
4 comments
4 participants
participants (4)
-
Allin Cottrell -
Artur T. -
Federico Fiorani -
Sven Schreiber