Just following up on this old email. On Fri, 2007-04-06 at 16:02 +1000, Brett Nash wrote: > > Despite TCP being reliable, it's fairly easy to corrupt a graphic > > download when using HTTP Resume. > > > > It is also useful when the modification time changes, but the contents > > doesn't. (It's also somewhat useful for finding duplicate media.) > > If you expect the mtime to change without changing the content, I'd > expect clients to be assuming the file as changed as well.
So you have never downloaded something over HTTP and found it to be corrupted? There are a number of ways for it to be corrupted and things like HTTP resume make it even worse. Then there are things like disk errors (which could be local or on a mirror server). What is wrong with having an extra check? > > It can't hurt to have, the client isn't forced to check the checksum. > > That is terrible logic. Why? My client which is going to use the checksum is going to always have the correct data. If your client can do it without using the checksum - good for you. It doesn't hurt you to have the checksums. Tim Ansell _______________________________________________ tp-devel mailing list [email protected] http://www.thousandparsec.net/tp/mailman.php/listinfo/tp-devel
