On Tue, 2006-03-14 at 12:53 -0500, James Ostheimer wrote:
> Oleg-
>
> I noticed that javadocs reccomend using getResponseBodyAsStream, and I
> changed my get code as follows:
>
...
>
> char[] cbuf = new char[1024];
> int len = 0;
> while((len = ir.read(cbuf)) > 0) {
> rawPage += new String(cbuf);
-----------------------------^
This is all horribly inefficient and wrong.
Try this (Disclaimer: the code may fail to compile. I just typed this
stuff directly in my mail client. Make changes you deem necessary)
char[] cbuf = new char[1024];
StringBuffer sbuf = new StringBuffer();
int len = 0;
while((len = ir.read(cbuf)) > 0) {
sbuf.append(cbuf, 0, len);
if (sbuf.length() > 100000) {
method.abort();
// Important. Do not forget to abort the
// method prior to releasing the connection
// Otherwise the connection manager will
// attempt to consume the remaining content
}
}
>
> So I now quit on any file over 100 kb.
No wonder.
> My memory still steadily rises and
> now Jprofiler claims:
>
> a.. 72.7% - 102,068 kB - 2,463 alloc. java.lang.StringBuffer.toString()
> from getPageApache(). That means that somewhere in the above code
> (including calls into HttpClient) I am eating up a lot of memory in
> StringBuffers.
>
No wonder.
Oleg
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
