On 18 October 2010 14:30, thanh nguyen <[email protected]> wrote: > Sorry for lack of precision. The random no are defined in the CSV file and > it's read at the beginning of script. JMeter doesn't create random no at > runtime. So I think it's more efficient?
No, because you are using Jexl to do the arithmetic; that will likely be less efficient. You could use http://jakarta.apache.org/jmeter/usermanual/component_reference.html#Constant_Timer with a value of ${__longSum(2000,${RANDOMNO})} Or fix the CSV file so it includes the 2000 offset and use ${RANDOMNO} But why not drop the CSV file and use http://jakarta.apache.org/jmeter/usermanual/component_reference.html#Uniform_Random_Timer instead? That will eliminate the CSV file processing as well as the arithmetic. > > Thank you > thanh > > On Mon, Oct 18, 2010 at 6:43 AM, sebb <[email protected]> wrote: > >> It is likely to be more efficient to use in-built Timers such as >> Random and Gaussian. >> >> Jexl, Beanshell, Javascript etc. all have to be interpreted first and >> then executed. Java code has already been interpreted. >> >> However whether this makes makes a significant difference to the >> running of a test is another matter. >> >> On 13 October 2010 16:25, thanh nguyen <[email protected]> wrote: >> > Hi all, >> > >> > I want to put some random think time in my script like this >> > : ${__jexl(2000+${RANDOMNO} ) }. But it doesn't work. 2000 alone, it is >> > working. I wonder why. Is it "cpu efficient" to use the object Gaussian >> > Random timer or Uniform Random Timer? >> > >> > Thank you, >> > >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: [email protected] >> For additional commands, e-mail: [email protected] >> >> > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
