|
Sort Order:
Addition log from selenium-server : 16:38:20.461 INFO - Command request: waitForPageToLoad[30000, ] on session befa3355d3df4c96b80db657aa7bf4b7 I suspect it happens because proxyInjectionMode adds another Frame , and the waitForLoad() method confuses this Frame with actual pages frames. We're having the same problem. Tried with the latest snapshot version as well and got the same problem. Using Selenium-RC with .NET. 11:39:03.533 INFO - Java: Sun Microsystems Inc. 14.0-b16 Date: Mar-3-2010 "java -jar selenium-server.jar -proxyInjectionMode -userContentTransformation old_test.js new_test.js" which I found 1.0.1 is the ONLY version capable to finish my unit test!(but such error msg. are still annoying) all other version simply exit before being able to act as a proxy. Please help! ERROR: I don't think that last one is strictly *piiexplore mode – with the latest IE8, on both Win7 and WinXP, I can get that last error to happen just by running a series of IE / HTA-mode tests and waiting for a bit. I haven't yet seen it in IE7, at least not within the same time frame. Here's the scenario: I have a 'hello world' test that uses the (default) *iexplore mode which launches the test in multiwindow mode, with IE in HTA mode. I have not yet tried to use a series of tests with *iexploreproxy. After 1-10 tests, the application window 'hello world' is not killed. No worries, RC doesn't seem to mind (although I'm rather unhappy about the browser window being open; what if there were temporary internet files associated with it that need to be removed?) and it keeps going. After about 5-25 more tests (each of these are sequential and a few minutes apart for ease of viewing) then the RC Remote Control window is the one that doesn't get killed although the associated 'hello world' window does – note that the first 'hello world' window is still there from the first problem. At that point, the command history shows 'getTitle()', which is the last selenium command executed before testComplete. Something is POSTing to the RC server (the RC control window?) for a really long period of time – as I type I'm only up to sequenceNumber 584, whereas the previous poster made it up to 264977. Sometimes, if I let this go on long enough, my winXP machine tells me it's rebooting because of ...well, I don't have a screenshot of that right now. Anyway, that of course 'fixes' the problem, until the next time I run the sequence. During this error state, if I then run various other tests (*iexploreproxy), they do pass. Until the machine can no longer start IE up fast enough, that is to say. Interestingly enough, once one of each type of window is left hanging open (control window or the 'hello world' window, the rate of abandoned windows drops quickly. It's been running now for 30 minutes and still just those two windows are there. There's a very good chance that the probability of hitting whatever race condition causes this drops quite a bit once the incessant POSTing start happening. I haven't checked in a while to see if *iexplore is actually now the same thing as *piiexplore, in which case this may be the same problem as initially reported. Otherwise, this may need to be made into a new bug. Ah, yes. It's when I send a shutDown() command to the flailing RC server that it reboots my machine. Note that I can kill the cmd window no problem - it's specifically when I call shutDown and then kill all the remaining iexplore tasks and it's been looping with the above errors. This could very well be a windows issue, but the fact remains that the windows cleanup sequence still needs some work. -Jen Update: I only see the looping POST warning above with IE8 – not with IE7. IE7 also has the same problem closing the windows, but does not result in anything continuing to send POSTS to the RC server. -Jen |
|||||||||||||||||||||||||||||||||||||
Bug
Open
Major
Proxy injection mode is not working
Correction, it does not happen with Opera on windows.