I was wondering, when a Gnutella query is sent out to all the clients, how does the client software find the file? i know you select the directory of shared files, but does the software catalog the files each time it's loaded, or does it just scan the folder(s) each time?

I would think that creating a database of files would take up more space, but make searching much faster.

There are functions to let you see the contents of directory. I'd imagin that client would get this information when it's load and store it in memory or a file and refreshes it from time to time. Then when it get's query for something like Modanna (Spelling?) it returns a query hit that contains all the matching files names (Any file containing the word Modanna) in the result list.

OK, so the client creates a list in memory or in a file and refreshes it manually.

i didnt think it actually scanned the folder each time.

for the exception of limeware (due to its non-native coding) most of the clients (morpheus, bearshare, etc..) load fairly quickly. I'm assuming each load of the program must create a new list, or atleast refresh the list. how do they get all the file attributes (name, size, etc..) so quickly?

They're not using VB!

Hah, i didnt think they were

They get the names with a routine that returns all files in a folder (involves and API call or two) or by having the user pick out the individual files to share. The sizes are easy, C\C++ and VB all have functions that can be used to get the file size. I am not sure what you mean by attribute, but if its the extension it's just a simple matter parsing the file name.