SolrJ would require some modification. SolrJ internally uses Jakarta HTTP Client via Solr's "CommonsHttpSolrServer" class. It would need to be ported to a different implementation of SolrServer (the base class), one that uses java.net.URL. I suggest "JavaNetUrlHttpSolrServer".
~ David Smiley On 4/14/09 1:13 PM, "Glen Newton" <glen.new...@gmail.com> wrote: I was wondering if those more up on SolrJ internals could take a look if there were any serious gotchas with the AppEngine's Java urlfetch with respect to SolrJ. http://code.google.com/appengine/docs/java/urlfetch/overview.html "The URL must use the standard ports for HTTP (80) and HTTPS (443). The port is implied by the scheme, but may also be mentioned in the URL as long as the port is standard for the scheme (https://...:443/). An app cannot connect to an arbitrary port of a remote host, nor can it use a non-standard port for a scheme." This is an annoyance for those running Solr on non-80/443. To some, this may be a fatal limitation. There is a 1M upload/download limit, which would impact large adds to the index and large results sets back from the index. There are also other quotas: http://code.google.com/appengine/docs/java/urlfetch/overview.html#Quotas_and_Limits Otherwise, my eyes see no other major issues. Others? thanks, Glen -- -
