Re[3]: log4j location information issue

[Ron Blaschke <ma...@rblasch.org>]

Any opinions on that one?


CG>> Since you are familiar with the relevant log4j code, do you know the stack
CG>> trace lines look like?

RB> Not an easy question.  In Java 1.4.2, Sun implements the information as either
RB> (<Filename>:<LineNumber>)
RB> (<Filename>)
RB> (Native Method)
RB> (Unknown Source)

RB> (see
RB> http://java.sun.com/j2se/1.4.2/docs/api/java/lang/StackTraceElement.html#toString())

RB> In 1.3.1 the implementation is similar, but directly in the VM (in
RB> method pc2string in classruntime.c).

RB> I'd recommend changing the code to the one below, which would leave
RB> "Native Method" and "Unknown Source" as is.  Replacing "Unknown Source"
RB> with NA might be considered, though.

RB>     public
RB>     String getFileName() {
RB>         if(fullInfo == null) return NA;

RB>         if(fileName == null) {
RB>             int iend = fullInfo.lastIndexOf(':');
RB>             if(iend == -1) {
RB>                 iend = fullInfo.lastIndexOf(')');
RB>             }

RB>             if (iend == -1) {
RB>                 fileName = NA;
RB>             } else {
RB>                 int ibegin = fullInfo.lastIndexOf('(', iend - 1);
RB>                 fileName = this.fullInfo.substring(ibegin + 1, iend);
RB>             }
RB>         }
RB>         return fileName;
RB>     }

Ron
-- 


---------------------------------------------------------------------
To unsubscribe, e-mail: log4j-dev-unsubscribe@jakarta.apache.org
For additional commands, e-mail: log4j-dev-help@jakarta.apache.org