Hi, is it possible that the regexp get executed in code called from a setTimeout / setInterval or XMLHttpRequest? In this case, with HtmlUnit-1.14 this occurs a separate thread and thread have a smaller heap allocated than the "main" thread. This would explain why you couldn't reproduce it. I guess that if you run your test in a separate thread you will be able to get the expected Error.
Cheers, Marc. -- Blog: http://mguillem.wordpress.com ruel loehr wrote: > Fixed by upping the stack size. I guess that section of the code has a > heavy footprint. > > > ruel loehr wrote: >> We are having some problems using the latest webtest release with dwr and >> javascript templates. >> >> JDK is 1.5.0.11. >> >> Initially, mark pointed us too some improvements made on htmlunit trunk >> for regex processing. >> >> I checked out html unit 1.14 tagged release and ported the trunk change >> for HTMLUnitRegexProxy to that branch. Everything worked fine for a bit, >> until our developers introduced dwr onto a second page. Our stack >> overflow errors then reappeared. >> >> [sleep] Exception in thread "HtmlUnit Managed Thread #7: >> XMLHttpRequest.send" java.lang.StackOverflowError >> [sleep] at >> java.util.regex.Pattern$Curly.match(Pattern.java:4190) >> >> I've hacked around at the htmlunit source, putting in a bunch of debug >> statements to try and narrow the problem. >> >> [sleep] INFO >> (com.gargoylesoftware.htmlunit.javascript.regexp.HtmlUnitRegExpProxy$RegExpData) >> - After escapeJSCurly re is ^(\s*\S*(\s+\S+)*)\s*$ >> [sleep] INFO >> (com.gargoylesoftware.htmlunit.javascript.regexp.HtmlUnitRegExpProxy) - >> Pattern compiled >> [sleep] INFO >> (com.gargoylesoftware.htmlunit.javascript.regexp.HtmlUnitRegExpProxy) - We >> are now calling pattern matcher with string: {var hourTypeIndicators = >> ["Branch Working Hours","Specific Hours","Not Available"]} >> [sleep] {var hourValues = >> ["01","02","03","04","05","06","07","08","09","10","11"]} >> [sleep] {var minuteValues = ["00","30"]} >> [sleep] {var ampmValues = ["AM","PM"]} >> [sleep] {var timeWindowSizes = [1, 2, 3,4, 5, 6,7, 8, 9,10, 11, 12,13, >> 14, 15,16,17, 18, 19,20, 21, 22,23,24]} >> [sleep] {if schedulingConfig != null} >> [sleep] {if (schedulingConfig.isBranchView)} >> >> I SNIPPED some the code which gets output here, it's a big javascript >> template. >> >> [sleep] >> [sleep] >> [sleep] INFO >> (com.gargoylesoftware.htmlunit.javascript.regexp.HtmlUnitRegExpProxy) - >> The matcher is created >> [sleep] INFO >> (com.gargoylesoftware.htmlunit.javascript.regexp.HtmlUnitRegExpProxy) - >> Calling replaceAll on the matcher with replacemetn : $1 >> [sleep] Exception in thread "HtmlUnit Managed Thread #7: >> XMLHttpRequest.send" java.lang.StackOverflowError >> [sleep] at >> java.util.regex.Pattern$Curly.match(Pattern.java:4190) >> >> Now the interesting part. I forced HTMLUnit to write out the the context >> string, the regex pattern and the flags to 3 respective files. Then I >> wrote a small test class to simulate the behavior. >> >> public static void main(String[] args) { >> // TODO Auto-generated method stub >> >> Regexp regexp = new Regexp("^(\\s*\\S*(\\s+\\S+)*)\\s*$"); >> >> String thisString= null; >> >> byte b[]; >> try { >> FileInputStream fis = new >> FileInputStream("c:\\string.txt"); >> int x= fis.available(); >> b = new byte[x]; >> >> fis.read(b); >> thisString = new String(b); >> >> } catch (FileNotFoundException e) { >> // TODO Auto-generated catch block >> e.printStackTrace(); >> } catch (IOException e) { >> // TODO Auto-generated catch block >> e.printStackTrace(); >> } >> >> System.out.println("About to compile a pattern"); >> System.out.println("Regexp is " + regexp.toString()); >> >> final Pattern pattern = Pattern.compile(regexp.toString(), 0 ); >> System.out.println("Pattern compiled"); >> final Matcher matcher = pattern.matcher(thisString); >> System.out.println("The matcher is created "); >> String done = matcher.replaceAll("$1"); >> >> } >> >> I can't get it to fail. Is the NativeRegExp class coming into play >> somehow and causing this bug? I haven't found a way to instantiate it in >> my test..... >> >> >> Thanks! >> Ruel Loehr >> >> >> >> >> >> > _______________________________________________ WebTest mailing list [email protected] http://lists.canoo.com/mailman/listinfo/webtest
