Re: Re[2]: [sqlite] Bug in the precision of strftime function
--- Slava Tutushkin <[EMAIL PROTECTED]> wrote: > > MJ> -4713-11-24 GC, which is -4712-01-01 JC" so datetime(0) does correctly > MJ> convert Julian Day 0 to a (proleptic) Gregorian date but not to a Julian > MJ> Date as (I think) I thought. That'll teach me to post follow-ups when > > I see no problem here while, because it does not matters, what day > julianday() references to, > all we need - stable pivot point. I'm using REAL julianday() value for > internal storage in > sqlite only, after reading from DB I'm converting to boost::ptime. > But strftime was not returning precise values sometimes. > > My post was about floating point problems with the milliseconds in the > strftime. I was > investigated it a little bit longer, and found out that even the query > "select strftime('%Y-%m-%d %H:%M:%f', '2006-09-24T10:50:26.046');" > is returning 2006-09-24 10:50:26.045. > > Anyway, somebody (thank you!) posted a patch in the ticket, solving this > problem. I was not > tested it for now, but seems it looks ok. > URL for ticket is: > http://www.sqlite.org/cvstrac/tktview?tn=1991 I had missed the boundary case for 59.+ seconds. Please use the latest version of the patch. __ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com - To unsubscribe, send email to [EMAIL PROTECTED] -
Re[2]: [sqlite] Bug in the precision of strftime function
MJ> -4713-11-24 GC, which is -4712-01-01 JC" so datetime(0) does correctly MJ> convert Julian Day 0 to a (proleptic) Gregorian date but not to a Julian MJ> Date as (I think) I thought. That'll teach me to post follow-ups when I see no problem here while, because it does not matters, what day julianday() references to, all we need - stable pivot point. I'm using REAL julianday() value for internal storage in sqlite only, after reading from DB I'm converting to boost::ptime. But strftime was not returning precise values sometimes. My post was about floating point problems with the milliseconds in the strftime. I was investigated it a little bit longer, and found out that even the query "select strftime('%Y-%m-%d %H:%M:%f', '2006-09-24T10:50:26.046');" is returning 2006-09-24 10:50:26.045. Anyway, somebody (thank you!) posted a patch in the ticket, solving this problem. I was not tested it for now, but seems it looks ok. URL for ticket is: http://www.sqlite.org/cvstrac/tktview?tn=1991 -- Slava Tutushkin, http://aloner.ru mailto:[EMAIL PROTECTED] ICQ: 55463183 - To unsubscribe, send email to [EMAIL PROTECTED] -