9:05 p.m.
Am 18.12.18 um 21:37 schrieb Allin Cottrell:
On Tue, 18 Dec 2018, Artur T. wrote:
> Hi all,
>
> the following leads t a crash using current git on ubuntu linux (both
> on 18.10 and 14.04) with the message
>
> <malloc(): memory corruption>
>
> <hansl>
> nulldata 14000
> series foo = 1
> strings S = array($nobs)
> loop i=1..$nobs -q
> S[i] = "Entry_$i"
> endloop
> stringify(foo,S)
> </hansl>
Thanks, Artur. That's now fixed in git.
Great, thank you for the quick fix, Allin.
Just one question for clarification. Running the following yields an --
at least for me -- unexpected behaviour where series foo results in a
constant ("Entry_c") even though S[i] is different for each i.
<hansl>
nulldata 14000
series foo = 1 # SET to 1
strings S = array($nobs)
loop i=1..$nobs -q
S[$i] = sprintf("Entry_%d", $i)
endloop
stringify(foo,S)
print S # results in "Entry_1" for all obs
series foo = 2 # SET to 2
strings S = array($nobs)
loop i=1..$nobs -q
S[$i] = sprintf("Entry_%d", $i)
endloop
print S # results in "Entry_2" for all obs
</hansl>
Best,
Artur
10:54 p.m.
New subject: Gretl crash with stringify() and big dataset
On Tue, 18 Dec 2018, Artur T. wrote:
Just one question for clarification. Running the following yields an
--
at least for me -- unexpected behaviour where series foo results in a
constant ("Entry_c") even though S[i] is different for each i.
<hansl>
nulldata 14000
series foo = 1 # SET to 1
strings S = array($nobs)
loop i=1..$nobs -q
S[$i] = sprintf("Entry_%d", $i)
endloop
stringify(foo,S)
print S # results in "Entry_1" for all obs
series foo = 2 # SET to 2
strings S = array($nobs)
loop i=1..$nobs -q
S[$i] = sprintf("Entry_%d", $i)
endloop
print S # results in "Entry_2" for all obs
</hansl>
Shouldn't really be unexpected.
Per stringify(), the strings from array @S are used only as needed
to encode the values of the series in question. Since the series you
define have only one value each (i.e. they are constant) they
require only one string, namely S[1].
Excess elements in the array of strings are tolerated, though if the
number of strings is deficient that provokes an error.
BTW, better not to use "$i" when "i" will do just fine. You need
"$i" only when you want the index value as a string, which is not
the case here.
Allin
11:11 p.m.
New subject: Gretl crash with stringify() and big dataset
On Tue, 18 Dec 2018, Allin Cottrell wrote:
On Tue, 18 Dec 2018, Artur T. wrote:
> Just one question for clarification. Running the following yields
> an -- at least for me -- unexpected behaviour where series foo
> results in a constant ("Entry_c") even though S[i] is different
> for each i.
>
> <hansl>
> nulldata 14000
> series foo = 1 # SET to 1
> strings S = array($nobs)
> loop i=1..$nobs -q
> S[$i] = sprintf("Entry_%d", $i)
> endloop
> stringify(foo,S)
> print S # results in "Entry_1" for all obs
>
> series foo = 2 # SET to 2
> strings S = array($nobs)
> loop i=1..$nobs -q
> S[$i] = sprintf("Entry_%d", $i)
> endloop
> print S # results in "Entry_2" for all obs
> </hansl>
Shouldn't really be unexpected.
Per stringify(), the strings from array @S are used only as needed
to encode the values of the series in question. Since the series
you define have only one value each (i.e. they are constant) they
require only one string, namely S[1].
Oops, not quite accurate: S[2] gets used in the second case. But
that's in line with the documentation in the chapter on
"String-valued series" in the User's Guide. Although a string-valued
series in principle represents an encoding that starts at 1, we
allow for the possibility that not all the integer values are
included in a finite sample.
Properly functioning example follows:
<hansl>
nulldata 20
series s = index
strings S = array($nobs)
loop i=1..$nobs -q
S[i] = sprintf("Value_%d", i)
endloop
stringify(s, S)
print s --byobs
</hansl>
Allin
12:01 p.m.
New subject: Gretl crash with stringify() and big dataset
Thanks for the clarification, Allin.
Best,
Artur
Am 19.12.18 um 00:11 schrieb Allin Cottrell:
On Tue, 18 Dec 2018, Allin Cottrell wrote:
> On Tue, 18 Dec 2018, Artur T. wrote:
>
>> Just one question for clarification. Running the following yields an
>> -- at least for me -- unexpected behaviour where series foo results
>> in a constant ("Entry_c") even though S[i] is different for each i.
>>
>> <hansl>
>> nulldata 14000
>> series foo = 1 # SET to 1
>> strings S = array($nobs)
>> loop i=1..$nobs -q
>> S[$i] = sprintf("Entry_%d", $i)
>> endloop
>> stringify(foo,S)
>> print S # results in "Entry_1" for all obs
>>
>> series foo = 2 # SET to 2
>> strings S = array($nobs)
>> loop i=1..$nobs -q
>> S[$i] = sprintf("Entry_%d", $i)
>> endloop
>> print S # results in "Entry_2" for all obs
>> </hansl>
>
> Shouldn't really be unexpected.
>
> Per stringify(), the strings from array @S are used only as needed to
> encode the values of the series in question. Since the series you
> define have only one value each (i.e. they are constant) they require
> only one string, namely S[1].
Oops, not quite accurate: S[2] gets used in the second case. But that's
in line with the documentation in the chapter on "String-valued series"
in the User's Guide. Although a string-valued series in principle
represents an encoding that starts at 1, we allow for the possibility
that not all the integer values are included in a finite sample.
Properly functioning example follows:
<hansl>
nulldata 20
series s = index
strings S = array($nobs)
loop i=1..$nobs -q
S[i] = sprintf("Value_%d", i)
endloop
stringify(s, S)
print s --byobs
</hansl>
Allin
_______________________________________________
Gretl-users mailing list
Gretl-users(a)lists.wfu.edu
http://lists.wfu.edu/mailman/listinfo/gretl-users
1943
days inactive
1950
days old
3 comments
2 participants
participants (2)
-
Allin Cottrell -
Artur T.