Conversion script: link fix

2015-02-26T23:48:56Z

link fix

New page

For a 377,715 character [[UniString]]?, this code takes over 3 minutes to run on my desktop:

<pre>
String mark = uString.get[[MarkCharacter]]( [[UniTokens]].FM );
String linesep = System.getProperty( "line.separator" );
uString.change( mark, linesep );
</pre>

Doing the same thing on the database server takes less than a second:

<pre>
OPEN 'HOLD' TO F.HOLD ELSE STOP
K.HOLD = 'FFAREP.01.WSMOAK.022004'
READ R.HOLD FROM F.HOLD, K.HOLD THEN
SWAP @FM WITH CHAR(010):CHAR(013) IN R.HOLD
END
</pre>

Admittedly my machine is less powerful than the 4-processor database server, but I still think 3 minutes is excessive. What is [[UniString]]? doing in there?!

I considered moving the string manipulation up to the database server and having a subroutine do the read and swap before returning the text, but thought I'd give Java a shot at it first:

<pre>
String temp = uString.toString();
String mark = uString.get[[MarkCharacter]]( [[UniTokens]].FM );
String linesep = System.getProperty( "line.separator" );
String temp2 = temp.replaceAll( mark, linesep );
</pre>

That runs in 1/3 of a second. Good news, I don't have to write a subroutine!

String.replaceAll arrived with JDK 1.4, so it's understandable that [[UniObjects]] for Java doesn't use it. However, until [[UniString]].change() is improved, I can't recommend using it for large Strings.

UniStringChange - Revision history

Conversion script: link fix