On 27 Mar 2012, at 3:42am, "Jay A. Kreibich" <j...@kreibi.ch> wrote:
> On Tue, Mar 27, 2012 at 03:30:03AM +0100, Simon Slavin scratched on the wall: >> >> On 27 Mar 2012, at 3:12am, Jay A. Kreibich <j...@kreibi.ch> wrote: >> >>> On Sun, Mar 25, 2012 at 05:48:01AM +0100, Simon Slavin scratched on the >>> wall: >>> >>>> Can those values be passed from a 'double' C variable ? I believe so. >>>> So I see no reason why SQLite shouldn't be storing them. >>> >>> If, in the sense of "can I arrange the bits that way", then yes. >>> However, in the sense of "is this a valid floating point number?" the >>> answer is no. >> >> Okay. That's the heart of the matter: what do we mean by "REAL" in the >> documentation. The word "REAL" right next to "IEEE" makes me think it >> handles NaN, +Inf and -Inf. But it's not explicitly stated. So fair >> point: if you don't claim to handle NaN then you don't have to. > > Not do drag things out, On /this/ list ? > but how would you "handle" a NaN? If someone > writes a signaling-NaN into the database, the DB will start to throw > floating point exceptions, terminating the host process, with almost > any calculation. Is that really something anyone would consider > "correct" behavior? Sounds more like a great attack vector to me. Okay, you inspired me to look up what special values exist in IEEE. I didn't remember many of them. According to Wikipedia you get no more than the following +0, -0 Subnormal numbers +Infinity, -Infinity Quiet and signaling NaN I think it might be reasonable to handle a signaling NaN at bind time -- to reject it before it even gets stored. Make up a new SQLITE_ result code which means 'attempt to bind unhandled value' which you could use for this and perhaps other things, like an attempt to bind a BLOB or zeroblob with negative length (unless there's already a result code for these, in which case use that for signaling NaNs). I think the OS's (or platform's) library routines can be used to handle everything else. They should give consistent results over all platforms. Simon. _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
