On Wed, 3 Mar 2004, Gilles Detillieux wrote: > According to Fred Stutzman: > > On Tue, 2 Mar 2004, Jim Cole wrote: > > > On Mar 2, 2004, at 5:30 PM, Fred Stutzman wrote: > > > > When I web search for a term that is indexed, the search returns > > > > results, > > > > everything is OK. When I web search for a term that is not indexed, I > > > > get > > > > an instantaneous 500 error. When I run htsearch from the command line, > > > > for terms indexed, it returns proper results. When I run htsearch > > > > from the > > > > command line for terms not indexed, it returns proper results. > > > > > > Have you modified the nomatch.html file in any way? Have you tried > > > temporarily removing that file or replacing it with trivial content > > > (e.g. opening and closing html/body tags)? How about setting the > > > nothing_found_file attribute to point to some entirely different > > > location? I think the code generally handles problems with this file > > > gracefully, but it might be worth a look since you have already ruled > > > out a lot of other possibilities. > > > > Hi there, > > > > I tried out your suggestion, unfortunately it didn't work. In the > > meantime, I discovered something quite odd. > > > > If you give the search a non-indexed term of 15 characters or more, it > > won't give you the 500 error. For example: > > > > http://iupac.org/cgi-bin/htsearch?config=htdig&restrict=&exclude=www.iupac.org%2Fgoldbook%2F&method=and&format=builtin-long&sort=score&words=859584758575867 > > > > however if you run that with less than 15 characters, you will get the 500 > > error: > > > > http://iupac.org/cgi-bin/htsearch?config=htdig&restrict=&exclude=www.iupac.org%2Fgoldbook%2F&method=and&format=builtin-long&sort=score&words=85958475857586 > > > > I'm really quite baffled, but I feel this might be a vital clue. > > Yes, and also quite odd is that it's not just the number of characters in > the words parameter that matters. If, for example, I remove the restrict= > parameter from your first URL above, I get the 500 error again, even with > a 15 character word or longer. It seems to be the total length of input, > or perhaps of output, that matters. > > The other oddity I noticed is that if I search for a word that works, like > for instance "iupac", I get results, but some or all of the page buttons > at the bottom are missing! When I view the page source of the results > page, it looks like the output is abrubtly truncated. However, it never > seems to get truncated mid-tag, so it doesn't seem to be a problem with > a buffer not being flushed. Something, though, is causing some of the > cout output from htsearch to be lost. > > > > Since you are installing from source, is upgrading to 3.1.6 an option? > > > I don't have any reason to be suspicious of the version you are using, > > > but 3.1.6 is the current official production release. It also addresses > > > a number of security issues. > > I second Jim's recommendation. However, you mention that you're running > on RedHat Enterprise Linux. I wonder if it might not be a bug in > libstdc++, with its ostream class. Have you installed all of Red Hat's > update RPMs for this system, especially any compiler/library updates? > What about apache updates? >
Gentlemen, I wanted to follow up on this, to provide a "solution" to others that might discover this thread. As Giles pointed out, the query string length was the key element in preventing 500 errors. I have hacked a solution by increasing the length of the query string. I no longer get 500 errors. This is still quite bizarre to me, but I'm very happy that I've found a way to eliminate the errors. Thanks, Fred > -- Fred Stutzman Desk: 962-5646 Cell: 260-8508 www.ibiblio.org ------------------------------------------------------- This SF.Net email is sponsored by: IBM Linux Tutorials Free Linux tutorial presented by Daniel Robbins, President and CEO of GenToo technologies. Learn everything from fundamentals to system administration.http://ads.osdn.com/?ad_id=1470&alloc_id=3638&op=click _______________________________________________ ht://Dig general mailing list: <[EMAIL PROTECTED]> ht://Dig FAQ: http://htdig.sourceforge.net/FAQ.html List information (subscribe/unsubscribe, etc.) https://lists.sourceforge.net/lists/listinfo/htdig-general
