Andreas Delmelle wrote:
If you would use FOP multiple times in a row, without restarting the JVM, then over a few runs that will save you minutes... The very first run is always a lot slower due to static initialization, class loading etc. Once the VM is warmed up, the average runtime for a formatting run will drastically reduce. To see what I mean, you could already make the comparison: try 50 isolated runs from the command-line, and afterwards, perform the same 50 runs, but then looped in a single small class.
Hmm. As a compromise, I may be able to create a java class that accepts multiple "commandlines" from stdin. This could be driven either via a pipe from a script (a sort of fop daemon) or via a command file, as per egrep --file (etc) I would prefer to avoid too much close-coupling with java in my particular environment (system intgration, coded in perl). In conjunction with my requirment to have variable size output, and 2 output formats, I think I need 3 runs in total, so reducing startup overhead seems a useful goal. BugBear --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
