sure you can, but first my example was very simplistic and second, you have additional work to maintain it.
Marc. Paul Rogers wrote: > well, youd record the script to see what the browser is sending for > the specified single user action. Then modify the jmeter recording to > put the changing data ( be it the users id, email address, book > purchase, phone number etc) into some form of variable. Since for > load on the server you can probably do enough of the logic in the > jmeter script to find out how well the server copes. > > Paul > > On Fri, Mar 14, 2008 at 12:55 PM, Marc Guillemot <[EMAIL PROTECTED]> wrote: >> Paul, >> >> you are missing the point that when more logic is handled on the client >> side, a session recording may just be wrong. Just imagine a client side >> logic to determine which page to load next based on the number of >> currently logged in clients. Load testing with the recorded session may >> be interesting, but it doesn't simulate the load that real users may >> produce. >> >> Additionally recording is a very bad solution in term of maintainability. >> >> Cheers, >> Marc. >> >> >> >> Paul Rogers wrote: >> > I see jmeter as a way to exercise the server. So even if your app is >> > heavily javascript, if you are using jmeter, then arent you only >> > interested in the server side aspects? So just record a session, and >> > those requests will be the server load based on the javascript usage. >> > Thats what Ive done for the app I work on , and its very heavy >> > javascript. I can see Im doing the right thing, as the server logs >> > from the load test, are the same as what I would expect from a user in >> > a browser. Or am I missing something? >> > >> > Paul >> > >> > On Fri, Mar 14, 2008 at 11:32 AM, Peter Loron <[EMAIL PROTECTED]> wrote: >> >> Having built-in Javascript support (and HTTPS proxy capture a la >> >> PureTest) would be a godsend for us. I'm working on doing some load >> >> testing of our web app, but it makes heavy use of HTTPS and >> >> Javascript. So far it is proving very hard to make the JMeter load >> >> match the "real world" activity. >> >> >> >> -Pete >> >> >> >> >> >> >> >> On Mar 14, 2008, at 7:25 AM, Marc Guillemot wrote: >> >> >> >> > I'm surely biased on this as I'm lead developer of HtmlUnit but I >> >> > really >> >> > think that a combination of JMeter & HtmlUnit (or WebTest) would bring >> >> > great possibilities in load testing of Ajax applications. >> >> > >> >> > HtmlUnit "is" a browser that evaluates the JS (nearly) like normal >> >> > browsers do. But it is so lightweight that it is possible to run a few >> >> > hundreds instances of WebClient (the "browser") in parallel on a >> >> > normal >> >> > computer. >> >> > >> >> > I don't know what the current status is with Dojo support, but complex >> >> > Ajax libraries are already supported by HtmlUnit and other will come. >> >> > >> >> > Cheers, >> >> > Marc. >> >> > -- >> >> > Blog: http://mguillem.wordpress.com >> >> > >> >> > >> >> > Woody Aichner wrote: >> >> >> Is anyone using Jmeter to test a web application that uses Web 2.0 >> >> >> capability and specifically one that uses DoJo. >> >> >> >> >> >> I have dealt with applications that make use of javascript before >> >> >> and do this by simulating what the javascript does in Jmeter. >> >> >> >> >> >> Now, with these heavily scripted applications, the job appears that >> >> >> it will be more difficult as the amount of javascript executed is >> >> >> very large. >> >> >> >> >> >> I realize that recording is one option, but have found that this >> >> >> does not work in alot of cases, because of the dynamic nature of >> >> >> the javascript. >> >> >> >> >> >> Woody --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
