AW: Problem with nrc 2.02 - string concatenation revisited.

[Thomas.Schneider]

Hello there,
   I am definitely sure that the Java compiler optimizes string
concatenation.
   For the same task I do need 3 seconds in (NetRexx ) and Java, and more
than 150 in Object Rexx.
  Simple benchmark results from my Rexx2Nrx work (bootstrapping my Rexx
Compiler)
hope this helps,
Tom, (and cheers from dark Vienna)
 

> -----Ursprüngliche Nachricht-----
> Von: Patric Bechtel [SMTP:[hidden email]]
> Gesendet am: Donnerstag, 6. Februar 2003 19:32
> An: Mike Cowlishaw
> Cc: [hidden email]
> Betreff: Re: Problem with nrc 2.02
>
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> On Thu, 6 Feb 2003 08:42:32 +0000, Mike Cowlishaw wrote:
>
> Hello Mike,
>
> >> options binary strictcase strictargs
> >> [...]
> >> outWriter=FileWriter(outfile)
> >> s=String
> >> [...]
> >> s='test;'id';'name';'count';'CRLF
> >> outWriter.write(char[] s.toCharArray())
>
> >The 'write' problem is almost unique in the Java libraries; I think you
> >should be able to achieve what you want by using:
>
> >  outWriter.write(String s)
>
> >though I would have expected the code you show to work.
>
> I tried that already. The first version of the code used all variations
> of "guessing" NetRexx to use the String version of write, so:
>
> s=String "whatsoever"
> outWriter.write(s)
>
> or even
>
> outWriter.write(String s)
>
> didn't work. I only thought it would NOT work around an array, so I
> tried the char[] variant, but no go.
>
> >On concatenation; I believe the Javac compiler will use the StringBuffer
> >class for concatenating strings -- check the bytecode listing.   Also,
> you
> >can avoid the NetRexx calls there if you make CRLF type String.
>
> phew. I'm really not sure. But even then, the annoyance of getting an NPE
> on concating a null String stays...
>
> tia
>
> Patric
>
>
> PGP Public Key Fingerprint: 2636 F26E F523 7D62  4377 D206 7C68 06BB
>
> -----BEGIN PGP SIGNATURE-----
> Version: PGPsdk version 1.7.1 (C) 1997-1999 Network Associates, Inc. and
> its affiliated companies.
>
> iQA+AwUBPkKcK3xoBrvMu8qQEQKygQCgrgdZVfVDCING5HihZK2i5JnNP/MAl393
> pdbvw4c//LvIeI+n4wq39b8=
> =8GvR
> -----END PGP SIGNATURE-----
>
>
> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
> ~~~
> To unsubscribe from this mailing list ( ibm-netrexx ), please send a note
> to
> [hidden email]
> with the following message in the body of the note
> unsubscribe ibm-netrexx <e-mail address>

[Patric Bechtel]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

On Thu, 6 Feb 2003 21:07:25 +0100, [hidden email] wrote:

Hallo Thomas, hello Mike,

I've just threw a closer look at the String concatenating issue (bytecode wise):

e.g. this java code:

    a="hallo"+i+"noch"+i+"mal";

translates to:

10 new #5 <Class java.lang.StringBuffer>
13 dup
14 invokespecial #6 <Method java.lang.StringBuffer()>
17 ldc #7 <String "hallo">
19 invokevirtual #8 <Method java.lang.StringBuffer append(java.lang.String)>
22 iload_2
23 invokevirtual #9 <Method java.lang.StringBuffer append(int)>
26 ldc #10 <String "noch">
28 invokevirtual #8 <Method java.lang.StringBuffer append(java.lang.String)>
31 iload_2
32 invokevirtual #9 <Method java.lang.StringBuffer append(int)>
35 ldc #11 <String "mal">
37 invokevirtual #8 <Method java.lang.StringBuffer append(java.lang.String)>
40 invokevirtual #12 <Method java.lang.String toString()>
43 astore_1

So my fears are, at least for the jdk1.4.1_01 compiler, away.

tia

Patric

>Hello there,
>   I am definitely sure that the Java compiler optimizes string
>concatenation.
>   For the same task I do need 3 seconds in (NetRexx ) and Java, and more
>than 150 in Object Rexx.
>  Simple benchmark results from my Rexx2Nrx work (bootstrapping my Rexx
>Compiler)
>hope this helps,
>Tom, (and cheers from dark Vienna)
>

>> -----Ursprüngliche Nachricht-----
>> Von: Patric Bechtel [SMTP:[hidden email]]
>> Gesendet am: Donnerstag, 6. Februar 2003 19:32
>> An: Mike Cowlishaw
>> Cc: [hidden email]
>> Betreff: Re: Problem with nrc 2.02
>>
>> -----BEGIN PGP SIGNED MESSAGE-----
>> Hash: SHA1
>>
>> On Thu, 6 Feb 2003 08:42:32 +0000, Mike Cowlishaw wrote:
>>
>> Hello Mike,
>>
>> >> options binary strictcase strictargs
>> >> [...]
>> >> outWriter=FileWriter(outfile)
>> >> s=String
>> >> [...]
>> >> s='test;'id';'name';'count';'CRLF
>> >> outWriter.write(char[] s.toCharArray())
>>
>> >The 'write' problem is almost unique in the Java libraries; I think you
>> >should be able to achieve what you want by using:
>>
>> >  outWriter.write(String s)
>>
>> >though I would have expected the code you show to work.
>>
>> I tried that already. The first version of the code used all variations
>> of "guessing" NetRexx to use the String version of write, so:
>>
>> s=String "whatsoever"
>> outWriter.write(s)
>>
>> or even
>>
>> outWriter.write(String s)
>>
>> didn't work. I only thought it would NOT work around an array, so I
>> tried the char[] variant, but no go.
>>
>> >On concatenation; I believe the Javac compiler will use the StringBuffer
>> >class for concatenating strings -- check the bytecode listing.   Also,
>> you
>> >can avoid the NetRexx calls there if you make CRLF type String.
>>
>> phew. I'm really not sure. But even then, the annoyance of getting an NPE
>> on concating a null String stays...
>>
>> tia
>>
>> Patric
>>
>>
>> PGP Public Key Fingerprint: 2636 F26E F523 7D62  4377 D206 7C68 06BB
>>
>> -----BEGIN PGP SIGNATURE-----
>> Version: PGPsdk version 1.7.1 (C) 1997-1999 Network Associates, Inc. and
>> its affiliated companies.
>>
>> iQA+AwUBPkKcK3xoBrvMu8qQEQKygQCgrgdZVfVDCING5HihZK2i5JnNP/MAl393
>> pdbvw4c//LvIeI+n4wq39b8=
>> =8GvR
>> -----END PGP SIGNATURE-----
>>
>>
>> ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>> ~~~
>> To unsubscribe from this mailing list ( ibm-netrexx ), please send a note
>> to
>> [hidden email]
>> with the following message in the body of the note
>> unsubscribe ibm-netrexx <e-mail address>
>~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
>To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to
>[hidden email]
>with the following message in the body of the note
>unsubscribe ibm-netrexx <e-mail address>

PGP Public Key Fingerprint: 2636 F26E F523 7D62  4377 D206 7C68 06BB

-----BEGIN PGP SIGNATURE-----
Version: PGPsdk version 1.7.1 (C) 1997-1999 Network Associates, Inc. and its affiliated companies.

iQA/AwUBPkK58nxoBrvMu8qQEQJMHQCggfBXLYq2KLWG0NRukOGZ6Or97O0AoP6z
/+dImOOqj3aSGwcSsNsFYndu
=vvDM
-----END PGP SIGNATURE-----




~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
To unsubscribe from this mailing list ( ibm-netrexx ), please send a note to
[hidden email]
with the following message in the body of the note
unsubscribe ibm-netrexx <e-mail address>