The RPC server unavailable error must be from when I manually close the IE window after the script has stalled out -- so far that I can tell, this is the only way to actually view the error, otherwise ruby/IE will hang out indefinitely. It can be ignored.
I did notice that I was able to "jump start" the application by clicking another link on the page. This also seems random/intermittent, however. If this other link is clicked sometimes the script picks up where it left off, usually causing any assertions you had for the correct page to fail, etc. Adam -----Original Message----- From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On Behalf Of Bret Pettichord Sent: Thursday, April 19, 2007 5:04 PM To: wtr-general@rubyforge.org Subject: Re: [Wtr-general] unknown property or method `readyState' Adam Reed wrote: > You know, I've been user the older dev version of watir and just > started receiving this error two days ago. Older scripts that were > working fine, all of a sudden stall out indefinitely, and give the > readyState error when I abort them. I just upgraded to 1165 today, > and still have the same problem. > > The page I am browsing uses no frames (all divs). I've also found > that a simple script such as one that goes to 10 different urls, with > no validations, assertions or other processing crashes after the > fourth link. It doesn't seem to matter what this fourth URL is (ie, > not a problem with the page itself, it can be anything). > > WIN32OLERuntimeError: Unknown property or method 'readyState' > HRESULT error code:0x800706ba > The RPC server is unavailable > ...watir.rb:1669:in 'method_missing' > ...watir.rb:1669:in 'wait' > ...watir.rb:1551:in 'goto' > > Any answers on this yet? > > Adam > My understanding it that this problem was introduced in build 1.5.1.1158. Previous build 1.5.1.1145 does not show this problem. Specifically I introduced it when i committed the recursive frames patch from bug 107 Here are the comments i made when i made this fix: Fix by Xavier Noria for http://jira.openqa.org/browse/WTR-107 I made a few additional changes to Xavier's patch: most dramatically i also removed the pre-existing rescue block, which did not make sense to me. I wish we had unit tests for this, but we don't. I decided that it was worth the risk... Well, clearly, the risk was more than i realized at the time. At the same time, if you are getting an RPC server unavailable error, then there is nothing i can do to fix it. To me, I have seen happen after you close an IE window. Unfortunately IE#close is asynchronous and thus unreliable. I don't use it personally, instead preferring to use IE#kill which is in contrib and currently untested. Maybe it is time to promote it. Could anyone who is getting these errors, please post the stack traces, like Adam. This is a hard problem to fix because it is intermittent and we can only really guess as to whether we really understand your problem. Bret _______________________________________________ Wtr-general mailing list Wtr-general@rubyforge.org http://rubyforge.org/mailman/listinfo/wtr-general _______________________________________________ Wtr-general mailing list Wtr-general@rubyforge.org http://rubyforge.org/mailman/listinfo/wtr-general
