On 01/10/2009, Milamber <[email protected]> wrote: > > > Le 30/09/2009 10:45, sebb a ecrit : > > > On 29/09/2009, Milamber <[email protected]> wrote: > > > > > > > Hello, > > > > > > Le 28/09/2009 01:54, sebb a ecrit : > > > > > > > > > > > > > On 27/09/2009, Milamber <[email protected]> wrote: > > > > > > > > > > > > > > > > > > > > > Hello, > > > > > > > > > > I works on French translation of new elements Comparison Assertion > and > > > > > Comparison Assertion Visualizer. I have some questions: > > > > > > > > > > * Some messages like "Response time:" aren't internationalized. I > would > > > > > like do this. Which resources files must be used? > > > > > - messages.properties > > > > > or > > > > > - CompareAssertionResources.properties > > > > > > > > > > > > > > > > > > > > > > > > CompareAssertionResources.properties is used for the > > > > > > > > > > > CompareAssertion GUI > > > > > > > > > > messages.properties is used for any GUIs which don't implement > TestBean. > > > > > > > > > > > > > > > > > > > For message "Response time:" in ComparisonAssertion class (which > implement > > > TestBean) but the message is only display in ComparisonVisualizer (which > not > > > implement TestBean), can you confirm the best resource file ? (I think > > > messages.properties...) > > > > > > > > > > Yes, messages.properties is used for any GUIs which don't implement > TestBean. > > > > Though of course for a French translation you need to edit > > messages_fr.properties. > > > > > Yes ;) > > > > If the GUI uses a fixed string rather than a resource property, then > > translation involves: > > - change code to use resource > > - add property to messages.properties > > - add translations to messages_xx.properties > > > > > > Ok, thanks for your instructions. > > > > > > > > > > > > > > > > > > > > > > > * In CompareAssertionResources.properties, I think > > > > > > > > > > > > > > > > > this > > > > > > > > > > > > > > > message isn't correct? > > > > > compareTime.shortDescription=Verify that all > Samplers' > > > > > return times are within a given number of milliseconds > > > > > > > > > > > > > > > > > > > > > > > > I think it needs to be qualified with "of each other". > > > > That could be inferred from that fact that the assertion compares > > > > samplers, but it would be better to be explicit. > > > > > > > > > > > > > > > > > > > > > > > > > In code, compare time behaviour verify that response time > difference > > > > > between previous and current are within a given number of ms. > > > > > (But a 'break instruction' used if compare time failure is found at > > > > > > > > > > > > > > > > > first > > > > > > > > > > > > > > > element under the controller, other aren't never tested - that's a > bug?) > > > > > > > > > > > > > > > > > > > > > > > > I don't think so - because the first failure wins. > > > > > > > > > > > > > > > > > > > With this test plan : > > > > > > -Thread Group (1-1-1) > > > |--Simple Controller > > > |--Loop controller (5 loops) > > > |--Java Sampler > > > |--Compare Assertion (compare content false and compare time with 100 > > > (ms)) > > > |--Comparison Assertion Visualizer > > > |--View Results Tree > > > > > > loop1: response time 271 ms > > > loop2: response time 295 ms => success in CAV and VRT (noting display > in > > > CAV, element tree is white) > > > loop3: response time 147 ms => failure in CAV and VRT (in CAV: left > pane: > > > RT 295 and right pane: RT 147, element tree is red) > > > loop4: response time 284 ms => failure in CAV and VRT (because loop3 > > > failed, no comparison with loop4 response time) (in CAV: left pane: RT > 295 > > > and right pane: RT 147 (not 284), element tree is red) > > > loop5: response time 275 ms => failure in CAV and VRT (because loop3 > > > failed, no comparison with loop5 response time) (in CAV: left pane: RT > 295 > > > and right pane: RT 147 (not 275), element tree is red) > > > > > > The "comparison assertion" mark as a failure all samplers within a > > > controller when the first test failed has been found. > > > Perhaps, it would be better to add a sentence in manual, because I was > in > > > mistake by the messages in Comparison Assertion Visualizer (always same > > > error message) > > > > > > Milamber > > > > > > > > > > The code was copied from an old Java 5 development - it looked as > > though it might be useful. Unfortunately, there is no documentation of > > how it is intended to work. > > > > I'm not sure now how I would expect a comparison assertion to work: > > - should the first response be treated as the "good" sample, and > > subsequent responses compared to it? > > > > > > That is this behaviour that I expect (that I understand in my mind when > when I talk comparison) > I made this changes in patch here (and some improvements) > https://issues.apache.org/bugzilla/show_bug.cgi?id=47907 > > The first is the 'good' sample and each sampler within controller is > compared to this first >
Yes, I think that's good. It has the advantage that one only needs to store one sample, instead of the list which is used at present. > > - should each response be compared with the previous response? If so, > > this would allow a steadily increasing (or decreasing) elapsed time > > without complaining. Also, if response 3 fails when compared with > > response 2, what should response 4 be compared against? > > > > > > No, I don't thought this... because, for compare response time, we can make > this with a Duration Assertion. > And Content comparison, with a regexp extractor and Response assertion. > > > > > If the behaviour needs to be changed to be more useful, now is the > > time to do it ... or maybe I should just remove the code again. > > > > > > With some little improvements, I think that will be a good new > functionality. In all cases, we can find some ways to do (ie dynamic time > comparison / content comparison), but with this assertion and this > visualizer, we have a speed way to do this. > > It's not very difficult to add a new field: Type Comparison: First against > Other(s) / Previous to Current / Mark on error at first failure > I can do this code changes, if you want. > > (type list messages have to rewrite in good English) > > Milamber > > > > --------------------------------------------------------------------- To unsubscribe, e-mail: [email protected] For additional commands, e-mail: [email protected]
