Re: Converting floats to ints with intval

On Thursday 06 May 2010 08:39:03 Bob McConnell wrote: > From: David McGlone > > >On Thursday 06 May 2010 07:19:48 Paul Waring wrote: > >> David Otton wrote: > >> > On 6 May 2010 11:52, Paul Waring <paul@xk7.net> wrote: > >> >> If I was designing the system from scratch, that's what I'd do. > >> >> Unfortunately this is an add-on to a legacy system where currency > > values > > >> >> are already stored as strings in the database (yes, not ideal I > > know, > > >> >> but you have to work with what you've got). > >> > > >> > I don't know much about your situation, but it does sound like you > >> > need to fix the root problem. I'd use a decimal type, and lean on > > the > > >> > database to do the maths. > >> > >> As I said, unfortunately it's a legacy system, so I can't just change > >> the database to use a different type (there are dozens of columns set > > up > > >> like this, with thousands of values already set). > > > > Would It be possible to write a script to extract everything from that > > > > database and insert it into a database with the correct columns and > > values, > > > then all you would need to do is change the db connection information. > > If the data is really stored in strings, you need to break it down into > substrings around the decimal and then convert both sides into integers > and combine them into an integer value. It is the conversion into float > that introduces the error because of the imprecise representation of > fractional decimal values in binary. I see. -- Blessings, David M.