Alejandro Henao González wrote:
I'm not sure that the problem is this.
And, many JSPs have this line.
Well, that would certainly multiply the problem, wouldn't it ?
Do you have any idea what these "system.gc()" were meant to achieve ?
Anyway, nothing that find and grep and a good text editor can't solve.
may be origin of problem is the hashmap static, because the method HTMLEncoder.encode not is static and the construct of HTMLEncoder initialize the object.
If you are not sure of your HTMLEncoder class, there are quite a few available that are
fully-tested.
A short search in Google for "java html encoder" shows a few, such as this one :
http://jtidy.sourceforge.net/multiproject/jtidyservlet/clover/org/w3c/tidy/servlet/util/HTMLEncode.html
(and there is another one in the Jakarta Commons).
Personally, instead of some kind of HashMap, I would just use a preset array of
String[128] with the translations of the first 128 char values (translating the inocuous
characters by themselves), and then translate everything else above 128 to "&#(char
value);". Intuitively, it would seem much faster than the complicated logic I see in
these classes (and yours).
But then, since I am no great Java expert, I may be missing something.
Getting back to the "system.gc()" however, I am pretty sure that if every access to a JSP
page on that system triggers a major GC, then there is no wonder about where your system's
CPU time is being spent.
That is also what Philip seemed to say in his comment to your thread dump. And he
definitely is a Java and Tomcat expert.
---------------------------------------------------------------------
To unsubscribe, e-mail: users-unsubscr...@tomcat.apache.org
For additional commands, e-mail: users-h...@tomcat.apache.org
