If you already have download candidates in your list of downloads then they should start to retry immediately. If you have no candidates Phex waits until you have a connection to the network and submits it's auto generated searches.


The automatic searches are now submitted in a more efficient way. That means there are less searches ( mostly unnecessary ) submitted.
The reason is that the mass of searches caused by Phex really plug up the network. This causes a overall lower performance of the network for ALL gnutella users. And other clients might block Phex to clean the network.
After a very complex analysis of the old automatic search I found out that most of the searches return the same results and that the research makes most sense when you are looking for rare files.
Also for rare files it makes no sense to submit a search every couple of seconds. You will not reach that many different hosts between two searches. Almost no gain for you but much hurt for the net. Therefore searches are submitted in bigger intervals.
If your network neighbourhood changes, you are opening up paths to new hosts therefore running searches are re-submitted to new connections once they are stable.

I hope you understand some of the design decisions we had to make. Phex also is a client that most users are running for a very long time ( 4 - 12 hours or longer ). For this kind of uptime it makes not much difference if a search is submitted every minute or in the average every 10 to 15 minutes. For users with lower uptime finding and downloading rare files is hopeless anyway.

Please tell me what you think.

Gregor