Rexx Strings vs. Rexx Indexed strings

[Thomas.Schneider.Wien]

  Hello NetRexx experts :-)

Is there a Rexx builtin method to decide *at run-time* whether a Rexx
String is an ordinary String or an INDEXED string?

Tom.
--
Thomas Schneider Projects ReyC & LOGOS on www.KENAI.com
_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Mike Cowlishaw]

 
> Is there a Rexx builtin method to decide *at run-time*
> whether a Rexx String is an ordinary String or an INDEXED string?

Not really -- because all Rexx objects (strings) can be indexed.

However, I suspect that looking at the owner or parent of a Rexx object
might tell you if it is part of a collection of strings -- but whether that
collection is itself a property of a Rexx object might be needed, too.  I
forget the implementation details ...

But possibly what you want is the exists() BIM?  This tells you if a given
index has been used.

Mike

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[David Requena]

Hmm, how about looping some variable over the Rexx instance. Its final value should tell you if there are any indexed sub-strings.

Not tested and probably not very efficient but it might be worth a try.

-
Saludos / Kind regards,
David Requena

-----Original Message-----
From: "Mike Cowlishaw" <[hidden email]>
Sender: [hidden email]
Date: Tue, 24 Aug 2010 10:56:27
To: <[hidden email]>; 'IBM Netrexx'<[hidden email]>
Reply-To: IBM Netrexx <[hidden email]>
Subject: RE: [Ibm-netrexx] Rexx Strings vs. Rexx Indexed strings

 
> Is there a Rexx builtin method to decide *at run-time*
> whether a Rexx String is an ordinary String or an INDEXED string?

Not really -- because all Rexx objects (strings) can be indexed.

However, I suspect that looking at the owner or parent of a Rexx object
might tell you if it is part of a collection of strings -- but whether that
collection is itself a property of a Rexx object might be needed, too.  I
forget the implementation details ...

But possibly what you want is the exists() BIM?  This tells you if a given
index has been used.

Mike

_______________________________________________
Ibm-netrexx mailing list
[hidden email]


_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Mike Cowlishaw]

 
> Hmm, how about looping some variable over the Rexx instance.
> Its final value should tell you if there are any indexed sub-strings.
>
> Not tested and probably not very efficient but it might be
> worth a try.

Yes that would work .. and could be used to count the number of indexed
strings.   Expensive for large sets of indices because it will always
generate the full set.

Just testing whether there are any there or not would be much simpler (but
would need a new one-line method).  But I'm not sure what that would be
useful for?

Mike

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Thomas.Schneider.Wien]

  Hi Mike & David,
     first I did reply to your first reply without reading the other msgs...

Second, I am just implementing some methods which will accept either
'standard Rexx' strings
*OR* indexed Rexx Strings as the arguments (for instance TITLE, PRINT,
DISPLAY etc).

I want to have them implicitely looping when the arguments are Indexed
Strings.

I first tried a lot to define a class RexxStem extends Rexx by my own
(you might remember the discussion years ago),
but then I cannot use the brackets anymore ...

Issue closed for the minute.

hope I didn't disturbe anybody...
Thomas.
======================================================================


Am 24.08.2010 12:54, schrieb Mike Cowlishaw:

>
>> Hmm, how about looping some variable over the Rexx instance.
>> Its final value should tell you if there are any indexed sub-strings.
>>
>> Not tested and probably not very efficient but it might be
>> worth a try.
> Yes that would work .. and could be used to count the number of indexed
> strings.   Expensive for large sets of indices because it will always
> generate the full set.
>
> Just testing whether there are any there or not would be much simpler (but
> would need a new one-line method).  But I'm not sure what that would be
> useful for?
>
> Mike
>
> _______________________________________________
> Ibm-netrexx mailing list
> [hidden email]
>
>

--
Thomas Schneider Projects ReyC & LOGOS on www.KENAI.com
_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Kermit Kiser]

Actually I often need to know if a Rexx "database" variable has indexed values or not . This is particularly true when trying to copy a multidimensional index (copyindexed only does one index level) or flattening the database variable for storage in external media. It might be nice to add a method to NetRexx to test for indexed values. I just ran the following test which seemed to work but I am not sure how efficient it would be with a large indexed collection. Are all indexes checked before the loop starts?

--------------------------------------------------------------
x="I am indexed"
x["a"]="a"
y="I am not indexed"

if indexed(x) then say "x is indexed"
              else say "x is not indexed"

if indexed(y) then say "y is indexed"
              else say "y is not indexed"                   
             
method indexed(v=Rexx) static returns boolean

    r=0
    loop i over v
        r=1
        leave
        end
       
    return r
---------------------------------------------------------------
results in:

x is indexed
y is not indexed


-- Kermit


On 8/24/2010 3:54 AM, Mike Cowlishaw wrote:

Hmm, how about looping some variable over the Rexx instance. 
Its final value should tell you if there are any indexed sub-strings. 

Not tested and probably not very efficient but it might be 
worth a try. 
    

Yes that would work .. and could be used to count the number of indexed
strings.   Expensive for large sets of indices because it will always
generate the full set.

Just testing whether there are any there or not would be much simpler (but
would need a new one-line method).  But I'm not sure what that would be
useful for?

Mike
  

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[David Requena]

Kermit,

Mike and me followed this discussion off-list inadvertently. Here is more or less what we came out with:

First:

    method isIndexed(str=Rexx) returns Rexx
        indexed = 0
        loop indexed over str
            indexed = 1
            leave
            end
        return indexed


Then:

    method isIndexed(str=Rexx) returns Rexx
        loop dummy over str
            return 1
             end
        return 0

Then:

    method isIndexed(str=Rexx) returns Rexx
        loop indexed over str
            indexed = 1
            return indexed
            end
        return 0

Then:

    method isIndexed(str=Rexx) returns Rexx
        loop indexed over str
            return indexed\=null
            end
        return 0

And finally:

    method isIndexed(str=Rexx) static returns Rexx
        loop str over str
            return 1
            end
         return 0

or:

    method isIndexed(str=Rexx) static returns Rexx
        loop str over str; return 1; end; return 0

The two latest being the same. Still not sure if this is right. It definitely works but
it's looping over indexed in theory no longer existing as str is 'assigned' to its first
index on entering the loop.

Mike is also concerned about efficiency for large collections of indexes. The initial snapshot
of all the indexes needs to be build for the LOOP OVER even when even when it will only loop
over the first (or none at all).

Now checking the spec. about the last implementation.

Is that it, Mike? I hope I didn't forget or misinterpret anything.


El 24/08/2010 21:05, Kermit Kiser escribió:

Actually I often need to know if a Rexx "database" variable has indexed values or not . This is particularly true when trying to copy a multidimensional index (copyindexed only does one index level) or flattening the database variable for storage in external media. It might be nice to add a method to NetRexx to test for indexed values. I just ran the following test which seemed to work but I am not sure how efficient it would be with a large indexed collection. Are all indexes checked before the loop starts?

--------------------------------------------------------------
x="I am indexed"
x["a"]="a"
y="I am not indexed"

if indexed(x) then say "x is indexed"
              else say "x is not indexed"

if indexed(y) then say "y is indexed"
              else say "y is not indexed"                   
             
method indexed(v=Rexx) static returns boolean

    r=0
    loop i over v
        r=1
        leave
        end
       
    return r
---------------------------------------------------------------
results in:

x is indexed
y is not indexed


-- Kermit


On 8/24/2010 3:54 AM, Mike Cowlishaw wrote:

Hmm, how about looping some variable over the Rexx instance. 
Its final value should tell you if there are any indexed sub-strings. 

Not tested and probably not very efficient but it might be 
worth a try. 
    

Yes that would work .. and could be used to count the number of indexed
strings.   Expensive for large sets of indices because it will always
generate the full set.

Just testing whether there are any there or not would be much simpler (but
would need a new one-line method).  But I'm not sure what that would be
useful for?

Mike
  

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Thomas.Schneider.Wien]

Hi David,
    thanks for the synopsis of what we can do currently.

I would suggest to use the function/method  isIndexed for the current minute (as described below, the variants are doing pretty the same job), and add the method IsIndexed to the Rexx class *as soon* as the source is opened.

As Mike did describe, a LOOP over an indexed string builds a snapshot of the indexed string at the start of the loop,
 i.e. does make a COPY of it, when I understand it right, so that the contents of str are not changed during the LOOP over....

Thus the loop may add members to the str, without affecting the current loop.

The same philosophy seems to be used in ooRexx, as far as I do know.

I can live with that for the minute, but we shall add it to class Rexx ASAP.

Thomas.
==================================================================================================
Am 24.08.2010 21:52, schrieb David Requena:

Kermit,

Mike and me followed this discussion off-list inadvertently. Here is more or less what we came out with:

First:

    method isIndexed(str=Rexx) returns Rexx
        indexed = 0
        loop indexed over str
            indexed = 1
            leave
            end
        return indexed


Then:

    method isIndexed(str=Rexx) returns Rexx
        loop dummy over str
            return 1
             end
        return 0

Then:

    method isIndexed(str=Rexx) returns Rexx
        loop indexed over str
            indexed = 1
            return indexed
            end
        return 0

Then:

    method isIndexed(str=Rexx) returns Rexx
        loop indexed over str
            return indexed\=null
            end
        return 0

And finally:

    method isIndexed(str=Rexx) static returns Rexx
        loop str over str
            return 1
            end
         return 0

or:

    method isIndexed(str=Rexx) static returns Rexx
        loop str over str; return 1; end; return 0

The two latest being the same. Still not sure if this is right. It definitely works but
it's looping over indexed in theory no longer existing as str is 'assigned' to its first
index on entering the loop.

Mike is also concerned about efficiency for large collections of indexes. The initial snapshot
of all the indexes needs to be build for the LOOP OVER even when even when it will only loop
over the first (or none at all).

Now checking the spec. about the last implementation.

Is that it, Mike? I hope I didn't forget or misinterpret anything.


El 24/08/2010 21:05, Kermit Kiser escribió:

Actually I often need to know if a Rexx "database" variable has indexed values or not . This is particularly true when trying to copy a multidimensional index (copyindexed only does one index level) or flattening the database variable for storage in external media. It might be nice to add a method to NetRexx to test for indexed values. I just ran the following test which seemed to work but I am not sure how efficient it would be with a large indexed collection. Are all indexes checked before the loop starts?

--------------------------------------------------------------
x="I am indexed"
x["a"]="a"
y="I am not indexed"

if indexed(x) then say "x is indexed"
              else say "x is not indexed"

if indexed(y) then say "y is indexed"
              else say "y is not indexed"                   
             
method indexed(v=Rexx) static returns boolean

    r=0
    loop i over v
        r=1
        leave
        end
       
    return r
---------------------------------------------------------------
results in:

x is indexed
y is not indexed


-- Kermit


On 8/24/2010 3:54 AM, Mike Cowlishaw wrote:

Hmm, how about looping some variable over the Rexx instance. 
Its final value should tell you if there are any indexed sub-strings. 

Not tested and probably not very efficient but it might be 
worth a try. 
    

Yes that would work .. and could be used to count the number of indexed
strings.   Expensive for large sets of indices because it will always
generate the full set.

Just testing whether there are any there or not would be much simpler (but
would need a new one-line method).  But I'm not sure what that would be
useful for?

Mike
  

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Mike Cowlishaw]

The two latest being the same. Still not sure if this is right. It definitely works but
it's looping over indexed in theory no longer existing as str is 'assigned' to its first
index on entering the loop.

Mike is also concerned about efficiency for large collections of indexes. The initial snapshot
of all the indexes needs to be build for the LOOP OVER even when even when it will only loop
over the first (or none at all).

Now checking the spec. about the last implementation.

Is that it, Mike? I hope I didn't forget or misinterpret anything.  

Yes that's it.  My concern about the last vesrion was that it could affect str in the caller -- I forget the parameter rules in Java/NetRexx.   No problem about whether the correct answer -- the collection (array?) of indices has to be built from str before the first one can be assigned to str. 

 

Mike

_______________________________________________
Ibm-netrexx mailing list
[hidden email]

[Mike Cowlishaw]

As Mike did describe, a LOOP over an indexed string builds a snapshot of the indexed string at the start of the loop,
 i.e. does make a COPY of it, when I understand it right, so that the contents of str are not changed during the LOOP over....
 

Loop Over takes a snapshot of the indices (if any) of the indexed string.  It does not make a copy of the string.

 

Mike

_______________________________________________
Ibm-netrexx mailing list
[hidden email]