At 09:22 AM 08/22/02 -0700, Rodney Barnett wrote:
>I'm using the prog method to index records from different tables and it's
>convenient to use the record ID as the file name along with a property
>indicating which table the record came from. So far it's working fine, but
>it occurred to me that the duplicate file names might cause some problems at
>some point especially if the incremental indexing proposed for 3.0 uses the
>file name as a unique key. Any thoughts?
My guess would be that there would be some way to say what property should
be used to check for existing files (so that they can be replaced) -- that
would be more flexible than requiring it to be always be the file name.
For example, a MD5 hash might be better for web docs that could be indexed
a / or /index.html.
That might be wrong, or that might need to be set during indexing because
there needs to be an index by whatever property is used to match files
Hum, I'm not really clear why the file name is even required when indexing
with the -S prog method.
Received on Thu Aug 22 16:37:53 2002