Hide Forgot
I have rpmfind-1.6-8. Here's what happens when I attempt to search for something that it can't find in the resource list: jik:~!124> rpmfind dnswalk rpmFetchDistribList: failed to grab metadata server list URI: http://www.redhat.com/RDF/resources/distribs/metadata.rdf Cannot install or locate resource dnswalk Do you want to search it in the catalog? [Y/n] : Y The index is 246 day(s) old, refetch [Y/n] : Y Loading catalog to /home/jik/.rpmfinddir/fullIndex.rdf.gz Searching the RPM catalog for dnswalk ... rdfRead /home/jik/.rpmfinddir/fullIndex.rdf.gz: resource is not wellformed XML Cannot open catalog /home/jik/.rpmfinddir/fullIndex.rdf.gz I loaded fullIndex.rdf.gz into an editor, and it looks fine.
Hum, this is likely due to either: - an http transfer problem or - an encoding problem If the file is not truncated the second is the more likely. Could you install the libxml2 package from rawhide and run : xmllint --noout /home/jik/.rpmfinddir/fullIndex.rdf.gz it will certify whether the file is well formed XML. If there is an error output could you provide the first few lines of the error so I get an idea of what is going on ? Also can you tell me what version of libxml you have installed ? rpm -q libxml thanks, Daniel
When I ran the xmllint command you asked me to run, I got this: /home/jik/.rpmfinddir/fullIndex.rdf.gz:28112: error: Extra content at the end of the document #f8wR:)b |<M&)>eHy FfGK;G7 | ^ When I zcat the fullIndex.rdf.gz, I see this: jik:~!135> zcat < /home/jik/.rpmfinddir/fullIndex.rdf.gz > /tmp/fullIndex.rdf zcat: stdin: decompression OK, trailing garbage ignored I'm sure these two errors are related to each other :-). It would appear that somehow fullIndex.rdf.gz is ending up with extra data at the end of it; I don't know whether that's a transfer problem or an actual corruption problem on the server. On the other hand, it would seem that rpmfind should be able to ignore this trailing garbage if zcat can. I have libxml-1.8.14-1.
Err, no sorry this is not a bug from libxml perspective. The XML spec is extremely clear about this, there is no end of stream marker and for the document to be well-formed (i.e. considered not corrupted) it must have the following content: http://www.w3.org/TR/REC-xml#NT-document with Misc being very precise concerning the characters accepted once the top element content is finished. http://www.w3.org/TR/REC-xml#NT-Misc So this extra garbage makes it a broken XML document, and hence it's not accessible to the rpmfind application. It really seems to be a transport error since I just fetched the fullIndex.rdf.gz here and checked, there is no error. This may be a bug in the HTTP client of libxml, if so it is likely to be fixed in libxml2. But it doesn't look like an rpmfind error so far, thanks for your feedback, Daniel