Encoding_OneByte would have to be changed, probably for the better. Buffers are created that would be stepped on in a multithreaded environment, and they aren't needed. The method encodeToCharset(String) should be doing the work in the result buffer that's created, and decodeFromCharset(byte[]) should just create one on the stack.

Oh, and the quick hack would need to be cleaned up:

private static Map encodingCache = new HashMap();
public static syncronized Encoding getEncoding(String encoding)
{
Encoding e;

if ( encoding == null )
{
encoding = defaultEncoding;
}

e = encodingCache.get(encoding);
if ( e == null )
{
e = createEncoding(encoding);
encodingCache.put(encoding, e);
}
return e;

//return createEncoding(encoding);
}

-----Original Message-----
From: Firebird-Java@yahoogroups.com [mailto:Firebird-Java@yahoogroups.com] On Behalf Of chiss14
Sent: Friday, June 02, 2006 3:26 PM
To: Firebird-Java@yahoogroups.com
Subject: [Firebird-Java] Possible memory leak in EncodingFactory ?


Hello everyone,

I ran into a problem using JayBird v2.0.1 with the embedded version of FireBird. It could very well be that I'm not using the right technique to query the db, I'm still quite new with this.

Anyway, a part of my application was doing a lot of plain SQL select and I was always running out of memory.

Using NetBeans's profiler, I found that there was close to 7000 instances of Encoding_cp1252 ... So I went to get the source code, found the beta 2.1.0 and traced my problem to the EncodingFactory class. It seems like its always returning new instances of Encoding_xxxx (which I guess in normal circumstances are getting cleaned up by the GC, but this wasn't happening in mine ;-) ).

So to make a long story short, I looked what the Encoding_xxx classes were doing, assumed they only perform "translation" work, i.e. could be reused easily. So I did the following quick hack :

private static HashMap encodingCache = new HashMap();

public static Encoding getEncoding(String encoding){

if (encoding == null) encoding = defaultEncoding;

if (!encodingCache.containsKey(encoding))
encodingCache.put(encoding,createEncoding(encoding));

return (Encoding)encodingCache.get(encoding);

//return createEncoding(encoding); }



...and this works like a charm! My (quite limited) tests looked ok, no adverse effects were observed. My memory usage is down (like, waaaaay down), which is all I wanted. Maybe this cache should go in a HashTable instead of an HashMap in a multithreaded context, I just don't know enough about JayBird's code.

Any thoughts ?

Francis



[Non-text portions of this message have been removed]





------------------------ Yahoo! Groups Sponsor --------------------~--> Home is just a click away.  Make Yahoo! your home page now.
http://us.click.yahoo.com/DHchtC/3FxNAA/yQLSAA/saFolB/TM
--------------------------------------------------------------------~->


Yahoo! Groups Links