On 29/06/16 09:45, Drago, William @ CSG - NARDA-MITEQ wrote:
> Aren't there things like that [checksums] already built in to the hard disk 
> controllers (CRC, Reed Solomon, etc.)?

They are at a different level and can only detect issues in what they
see.  For example SQLite can create a page of data, and then hand it off
to the C library which then hands it off to the kernel which then hands
it off to various filesystem drivers which then hand it off to various
block devices which then hand it off over a bus of some sort to the
storage.  If corruption happens at any point before getting to the
storage then the corrupted version is going to be considered correct by
the storage.

Having checksums at the SQLite level means that SQLite can itself verify
that what it wrote (and went through any number of other layers) is what
it gets back.  Short of extremely robust C libraries, operating systems,
drivers, and hardware, SQLite is the sensible place to add checksums.
The "Lite" bit guarantees that SQLite is not run on robust everything,
but usually on less reliable components.  That is why I am somewhat
disappointed the SQLite team doesn't see value in implementing the request.

Roger


Attachment: signature.asc
Description: OpenPGP digital signature

_______________________________________________
sqlite-users mailing list
sqlite-users@mailinglists.sqlite.org
http://mailinglists.sqlite.org/cgi-bin/mailman/listinfo/sqlite-users

Reply via email to