Re: Odd [math] behavior

This WebDNA talk-list message is from

2003


It keeps the original formatting.
numero = 52784
interpreted = N
texte = What if you allowed a Format specification in the Math context? [math format=1.2f]17/7[/math] This could be combined with an OPTIONAL default precision pref, but I wouldn't want an upgrade installation to break any of my existing math, some of which is pretty complicated and needs a high precision. - brian On Thursday, September 11, 2003, at 01:15 PM, Scott Anderson wrote: > > I cringed when I saw this thread, knowing that it would lead to a > discussion > of the inaccuracies of binary representation of decimal numbers. > > I got a kick out of the following quote (taken from one of the links > John > provided - Thanks, John!). > > "Floating point is by its nature inexact. It is probably best if you > imagined that after every floating point operation, a little demon > came in > and added or subtracted a tiny number to fuzz the low order bits of > your > result..." > > I am sure that even the most experienced programmer has gotten 'bit' by > this. > > So, I think the lesson here is to ALWAYS keep in mind the desired > 'precision' when performing floating point arithmetic and use the > [format > x.xf] to ensure that you get what you expect. > > > FEEDBACK REQUEST: Given that most WebDNA programmers would not expect > to > have to deal with those 'fuzzy' low order bits, perhaps we should > build in a > more user-friendly default precision for the MATH context. Maybe have > it > controlled via a new WebDNA pref? Any thoughts? > > > > > > > -----Original Message----- > From: John Peacock [mailto:jpeacock@rowman.com] > Sent: Thursday, September 11, 2003 12:23 PM > To: WebDNA-Talk@talk.smithmicro.com > Subject: Re: Odd [math] behaviour > > > Paul Willis wrote: > >> I fully agree, I don't understand why any calculations or formatting > are >> required. This is simple subtraction. 6000.4-5300.3 is 700.1 a small >> child could do it. >> > > Unfortunately, computers are far stupider than even the smallest child. > The > advantage of using a computer over a child is that the computer > actually > does > what you tell it to do and does it quickly (no "In a minute Daddy!"). > > A quick Google yielded a few useful pages: > > http://www.python.org/doc/current/tut/node14.html > http://mindprod.com/jgloss/floatingpoint.html > http://www.rwc.uc.edu/koehler/comath/14.html > > HTH > > John ------------------------------------------------------------- This message is sent to you because you are subscribed to the mailing list . To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://webdna.smithmicro.com/ Associated Messages, from the most recent to the oldest:

    
  1. Re: Odd [math] behavior ( Brian Fries 2003)
  2. Re: Odd [math] behavior ( Glenn Busbin 2003)
  3. Re: Odd [math] behavior ( "Scott Anderson" 2003)
  4. Re: Odd [math] behavior ( John Peacock 2003)
  5. Re: Odd [math] behavior ( Brian Fries 2003)
  6. Re: Odd [math] behavior ( Chris List Recipient 2003)
  7. Re: Odd [math] behavior ( "Scott Anderson" 2003)
What if you allowed a Format specification in the Math context? [math format=1.2f]17/7[/math] This could be combined with an OPTIONAL default precision pref, but I wouldn't want an upgrade installation to break any of my existing math, some of which is pretty complicated and needs a high precision. - brian On Thursday, September 11, 2003, at 01:15 PM, Scott Anderson wrote: > > I cringed when I saw this thread, knowing that it would lead to a > discussion > of the inaccuracies of binary representation of decimal numbers. > > I got a kick out of the following quote (taken from one of the links > John > provided - Thanks, John!). > > "Floating point is by its nature inexact. It is probably best if you > imagined that after every floating point operation, a little demon > came in > and added or subtracted a tiny number to fuzz the low order bits of > your > result..." > > I am sure that even the most experienced programmer has gotten 'bit' by > this. > > So, I think the lesson here is to ALWAYS keep in mind the desired > 'precision' when performing floating point arithmetic and use the > [format > x.xf] to ensure that you get what you expect. > > > FEEDBACK REQUEST: Given that most WebDNA programmers would not expect > to > have to deal with those 'fuzzy' low order bits, perhaps we should > build in a > more user-friendly default precision for the MATH context. Maybe have > it > controlled via a new WebDNA pref? Any thoughts? > > > > > > > -----Original Message----- > From: John Peacock [mailto:jpeacock@rowman.com] > Sent: Thursday, September 11, 2003 12:23 PM > To: WebDNA-Talk@talk.smithmicro.com > Subject: Re: Odd [math] behaviour > > > Paul Willis wrote: > >> I fully agree, I don't understand why any calculations or formatting > are >> required. This is simple subtraction. 6000.4-5300.3 is 700.1 a small >> child could do it. >> > > Unfortunately, computers are far stupider than even the smallest child. > The > advantage of using a computer over a child is that the computer > actually > does > what you tell it to do and does it quickly (no "In a minute Daddy!"). > > A quick Google yielded a few useful pages: > > http://www.python.org/doc/current/tut/node14.html > http://mindprod.com/jgloss/floatingpoint.html > http://www.rwc.uc.edu/koehler/comath/14.html > > HTH > > John ------------------------------------------------------------- This message is sent to you because you are subscribed to the mailing list . To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://webdna.smithmicro.com/ Brian Fries

DOWNLOAD WEBDNA NOW!

Top Articles:

Talk List

The WebDNA community talk-list is the best place to get some help: several hundred extremely proficient programmers with an excellent knowledge of WebDNA and an excellent spirit will deliver all the tips and tricks you can imagine...

Related Readings:

NetForms for email (1998) Date Formats (1997) redirect with frames (1997) Upload files script (2003) WebDNA - fun to compare (2005) [TaxableTotal] - not working with AOL and IE (1997) Problems problems problems (1997) [OT] (waaaay OT) further off topic than before (2004) Upgrade to WebCat2 from Commerce Lite (1997) back button loses cart (2000) credit card question (simple) (2001) WebCatalog can't find database (1997) WebCat2 as a chat server? (1997) One other big addition... (1997) searching for items that begin with a number (2004) searchable list archive (1997) Emailer [cart] file names (1997) WebCatalog 2.0 b 15 mac (1997) Projects & Contractors (1997) WebCat2b13MacPlugIn - More limits on [include] (1997)