[ast-developers] Double Associative arrays patch help request

[Mohan Gorai]

Hi ,

I'm using to ksh93 scripts in my application to do various jobs . while
using double associative arrays i found the below :

# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )

<<== foo[baz][0] is inserted unexpectedly

Found that it's being fixed in alpha release as part of the below

12-08-20 A bug in which creating a two dimemsional associative array could
add an extra 0 element to the second subscript has been fixed.

could someone help me the patch file so that i can patch my binary
accordingly .


Appreciate any help here .

Thanks!,
~gmr
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131114/3cefc57c/attachment.html>

[Joshuah Hurst]

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do various jobs . while
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of the below
12-08-20 A bug in which creating a two dimemsional associative array could
add an extra 0 element to the second subscript has been fixed.
could someone help me the patch file so that i can patch my binary
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If you wish to do
the same for associative arrays just append the strings.

The 12-08-20 issue you're referring to was a bug which tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not defined variables
yet and set -o nounset was not set.

Arguably the right fix here is that ksh93 should throw an error if
someone tried to use an associative array with more than one
dimension.

Josh

[Mohan Gorai]

Hi Josh,

Thanks for the response.

From the output below, I see that we do have the arrary elements with the

strings as index created. But, an extra array member is being created for
every multidimensional associative array.

$ echo "${.sh.version}"
Version JM 93u 2011-02-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([0]='' [cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
$


But with the alpha release, I don't see the extra element.

$ echo "${.sh.version}"
Version AIJMP 93v- 2013-10-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([fish]=4 [mouse]=3) )
$

So it looks to me that ksh93 does support the associative arrays with more
than one dimension.

Am I missing something here?

Thanks,
~gmr

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do various jobs . while
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of the below
12-08-20 A bug in which creating a two dimemsional associative array

could

Post by Mohan Gorai
add an extra 0 element to the second subscript has been fixed.
could someone help me the patch file so that i can patch my binary
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If you wish to do
the same for associative arrays just append the strings.
The 12-08-20 issue you're referring to was a bug which tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not defined variables
yet and set -o nounset was not set.
Arguably the right fix here is that ksh93 should throw an error if
someone tried to use an associative array with more than one
dimension.
Josh

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131114/c55d798d/attachment.html>

[Rohith]

Hi Mohan,

I have attached the patch file .

Hope it helps .

Thanks
Rohith

Post by Mohan Gorai
Hi Josh,
Thanks for the response.
From the output below, I see that we do have the arrary elements with
the strings as index created. But, an extra array member is being
created for every multidimensional associative array.
$ echo "${.sh.version}"
Version JM 93u 2011-02-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([0]='' [cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
$
But with the alpha release, I don't see the extra element.
$ echo "${.sh.version}"
Version AIJMP 93v- 2013-10-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([fish]=4 [mouse]=3) )
$
So it looks to me that ksh93 does support the associative arrays with
more than one dimension.
Am I missing something here?
Thanks,
~gmr
On Thu, Nov 14, 2013 at 3:16 PM, Joshuah Hurst <joshhurst at gmail.com
On Thu, Nov 14, 2013 at 8:26 AM, Mohan Gorai <mohangorai at gmail.com

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do various jobs

. while

Post by Mohan Gorai
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of the below
12-08-20 A bug in which creating a two dimemsional associative

array could

Post by Mohan Gorai
add an extra 0 element to the second subscript has been fixed.
could someone help me the patch file so that i can patch my binary
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If you wish to do
the same for associative arrays just append the strings.
The 12-08-20 issue you're referring to was a bug which tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not defined variables
yet and set -o nounset was not set.
Arguably the right fix here is that ksh93 should throw an error if
someone tried to use an associative array with more than one
dimension.
Josh
_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131116/efa4a0c9/attachment.html>
-------------- next part --------------
--- a/src/cmd/ksh93/sh/array.c 2013-11-15 16:48:32.816237858 +0530
+++ b/src/cmd/ksh93/sh/array.c 2013-11-15 16:47:26.360240354 +0530
@@ -1003,7 +1003,7 @@
ap->nelem = nelem;
ap->fun = fun;
nv_onattr(np,NV_ARRAY);
- if(fp || value)
+ if(fp || (value && value!=Empty))
{
nv_putsub(np, "0", ARRAY_ADD);
if(value)

[David Korn]

That is the correct patch. Thanks.

Post by Rohith
Hi Mohan,
I have attached the patch file .
Hope it helps .
Thanks
Rohith
Hi Josh,
Thanks for the response.
From the output below, I see that we do have the arrary elements with the
strings as index created. But, an extra array member is being created for
every multidimensional associative array.
$ echo "${.sh.version}"
Version JM 93u 2011-02-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([0]='' [cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
$
But with the alpha release, I don't see the extra element.
$ echo "${.sh.version}"
Version AIJMP 93v- 2013-10-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([fish]=4 [mouse]=3) )
$
So it looks to me that ksh93 does support the associative arrays with more
than one dimension.
Am I missing something here?
Thanks,
~gmr

Post by Joshuah Hurst

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do various jobs . while
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of the below
12-08-20 A bug in which creating a two dimemsional associative array

could

Post by Mohan Gorai
add an extra 0 element to the second subscript has been fixed.
could someone help me the patch file so that i can patch my binary
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If you wish to do
the same for associative arrays just append the strings.
The 12-08-20 issue you're referring to was a bug which tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not defined variables
yet and set -o nounset was not set.
Arguably the right fix here is that ksh93 should throw an error if
someone tried to use an associative array with more than one
dimension.
Josh

_______________________________________________
ast-developers mailing listast-developers at lists.research.att.comhttp://lists.research.att.com/mailman/listinfo/ast-developers
_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131115/e469821f/attachment.html>

[Mohan Gorai]

Thanks Rohith / David for the patch.

cheers,
~gmr

Post by David Korn
That is the correct patch. Thanks.

Post by Rohith
Hi Mohan,
I have attached the patch file .
Hope it helps .
Thanks
Rohith
Hi Josh,
Thanks for the response.
From the output below, I see that we do have the arrary elements with
the strings as index created. But, an extra array member is being created
for every multidimensional associative array.
$ echo "${.sh.version}"
Version JM 93u 2011-02-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([0]='' [cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
$
But with the alpha release, I don't see the extra element.
$ echo "${.sh.version}"
Version AIJMP 93v- 2013-10-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([fish]=4 [mouse]=3) )
$
So it looks to me that ksh93 does support the associative arrays with
more than one dimension.
Am I missing something here?
Thanks,
~gmr

Post by Joshuah Hurst

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do various jobs . while
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of the below
12-08-20 A bug in which creating a two dimemsional associative array

could

Post by Mohan Gorai
add an extra 0 element to the second subscript has been fixed.
could someone help me the patch file so that i can patch my binary
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If you wish to do
the same for associative arrays just append the strings.
The 12-08-20 issue you're referring to was a bug which tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not defined variables
yet and set -o nounset was not set.
Arguably the right fix here is that ksh93 should throw an error if
someone tried to use an associative array with more than one
dimension.
Josh

_______________________________________________
ast-developers mailing listast-developers at lists.research.att.comhttp://lists.research.att.com/mailman/listinfo/ast-developers
_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com
http://lists.research.att.com/mailman/listinfo/ast-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131116/4fccfa59/attachment-0001.html>

[Rohith]

Hi David ,

I was curious to know . Generally is there a place where one can find
the patches for the relevant bugs ?
Do we maintain it somewhere ?

Thanks In Advance
Rohith

Post by Mohan Gorai
Thanks Rohith / David for the patch.
cheers,
~gmr
On Sat, Nov 16, 2013 at 3:10 AM, David Korn <dgkorn at gmail.com
That is the correct patch. Thanks.
On Fri, Nov 15, 2013 at 1:55 PM, Rohith <rohith.ramesh at oracle.com
Hi Mohan,
I have attached the patch file .
Hope it helps .
Thanks
Rohith

Post by Mohan Gorai
Hi Josh,
Thanks for the response.
From the output below, I see that we do have the arrary
elements with the strings as index created. But, an extra
array member is being created for every multidimensional
associative array.
$ echo "${.sh.version}"
Version JM 93u 2011-02-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([0]='' [cat]=1 [dog]=2) [baz]=([0]='' [fish]=4 [mouse]=3) )
$
But with the alpha release, I don't see the extra element.
$ echo "${.sh.version}"
Version AIJMP 93v- 2013-10-08
$ typeset -A foo
$ typeset -A foo[bar]
$ typeset -A foo[baz]
$ foo[bar][cat]=1
$ foo[bar][dog]=2
$ foo[baz][mouse]=3
$ foo[baz][fish]=4
$ set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([fish]=4 [mouse]=3) )
$
So it looks to me that ksh93 does support the associative
arrays with more than one dimension.
Am I missing something here?
Thanks,
~gmr
On Thu, Nov 14, 2013 at 3:16 PM, Joshuah Hurst
On Thu, Nov 14, 2013 at 8:26 AM, Mohan Gorai

Post by Mohan Gorai
Hi ,
I'm using to ksh93 scripts in my application to do

various jobs . while

Post by Mohan Gorai
# typeset -A foo
# typeset -A foo[bar]
# typeset -A foo[baz]
# foo[bar][cat]=1
# foo[bar][dog]=2
# foo[baz][mouse]=3
# foo[baz][fish]=4
# set | fgrep foo
_='foo[baz]'
foo=([bar]=([cat]=1 [dog]=2) [baz]=([0]='' [fish]=4

[mouse]=3) )

Post by Mohan Gorai
<<== foo[baz][0] is inserted unexpectedly
Found that it's being fixed in alpha release as part of

the below

Post by Mohan Gorai
12-08-20 A bug in which creating a two dimemsional

associative array could

Post by Mohan Gorai
add an extra 0 element to the second subscript has

been fixed.

Post by Mohan Gorai
could someone help me the patch file so that i can

patch my binary

Post by Mohan Gorai
accordingly .

ksh93 does not support multidimensional associative arrays.
Only indexed multidimensional arrays are supported. If
you wish to do
the same for associative arrays just append the strings.
The 12-08-20 issue you're referring to was a bug which
tried to create
multidimensional indexed arrays using string as index, which are
equivalent to the arithmetic value 0 if they are not
defined variables
yet and set -o nounset was not set.
Arguably the right fix here is that ksh93 should throw an
error if
someone tried to use an associative array with more than one
dimension.
Josh
_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com <mailto:ast-developers at lists.research.att.com>
http://lists.research.att.com/mailman/listinfo/ast-developers

_______________________________________________
ast-developers mailing list
ast-developers at lists.research.att.com
<mailto:ast-developers at lists.research.att.com>
http://lists.research.att.com/mailman/listinfo/ast-developers

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.research.att.com/pipermail/ast-developers/attachments/20131118/d0050635/attachment-0001.html>