On Thu, Oct 16, 2008 at 08:38:33AM -0700, Brian Aker wrote: > On Oct 16, 2008, at 5:46 AM, Andrey Hristov wrote: > >> forgot to add something. In the MySQL C/S protocol we are missing a >> check-sum to check whether the data is correct. I am aware that TCP >> has own check-summing but it protects only from problems during the >> network transmission, not memory problems. As this is not often the >> case it should be pluggable. As the server cannot know the CRC when >> starting to send the data to the client, it should be sent as last > > I like the idea of optionally sending this. We know this is needed in > replication and in general its not a bad idea to allow for this > optionally.
I think we should checksum individual bits of data, not giant chunks. For each batch of rows sent back, checksum that. Not for the entire result set. Remember that for the checksum to be useful we have to be able to know it's bad *before* processing any of the data. Considering that the code to compute small checksums is tiny, why not just have it always on? -- Stewart Smith _______________________________________________ Mailing list: https://launchpad.net/~drizzle-discuss Post to : [email protected] Unsubscribe : https://launchpad.net/~drizzle-discuss More help : https://help.launchpad.net/ListHelp
