Re: [Nutch-dev] inject breaks with "java.lang.StackOverflowError"

2005-02-27 Thread Michael Nebel
Hi, I found the following hack for my problem, but I think, someone with a better understanding of the code should take a look at this case +++ WebDBWriter.java2005-02-27 21:38:08.0 +0100 --- WebDBWriter.java.~1.52.~2004-12-05 10:43:42.0 +0100 @@ -533,24 +532,16 @

Re: [Nutch-dev] inject breaks with "java.lang.StackOverflowError"

2005-02-27 Thread Michael Nebel
Hi, I think, I hunted the problem a little bit down: WebDBWriter.java ... // Create a brand-new output db for the integrated data MapFile.Writer newDb = (comparator == null) ? new MapFile.Writer(fs, newDbFile.getPath(), keyClass, valueClass) : new MapFile.Writer(fs, newDbFile.getPath(), comp

[Nutch-dev] inject breaks with "java.lang.StackOverflowError"

2005-02-27 Thread Michael Nebel
Hi, I checked out the latest sources from the CVS and ran ant. Then I created a new webdb and injected a url. This job breaks with a StackOverflowError (see below). I looking at the code for about an hour, but I don't have any idea, what's going wrong. Does anybody have a hint for me? I'm a lit

Re: [Nutch-dev] search suggestion

2005-02-27 Thread Matt Kangas
So, you're suggesting to integrate user-feedback into a site/page's ranking? One might argue that Google already does this quite well, by scouring their httpd logs for signs of user dissatisfaction, such as needing to click past the first page. http://www.google.com/search?q=google+%22first+page%