- Pharo-dev - lists.pharo.org

Re: [Pharo-dev] Division isn't correct

by stepharo

I suggest you to read the Small number chapter of the Deep into Pharo. Stef On 11/7/14 15:53, Natalia Tymchuk wrote: > Hello. > I found interesting thing: > Why it is like this? > > Best regards, > Natalia >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Nicolas Cellier

2014-07-11 17:14 GMT+02:00 Camille Teruel <camille.teruel(a)gmail.com>: > > On 11 juil. 2014, at 17:02, Natalia Tymchuk <natalia.tymchuk(a)unikernel.net> > wrote: > > But why > 185/10 = 18.5 -> true? > > > Because 18.5 has an exact float representation, while 0.2 hasn't. > > Couldn't we use #asApproximateFraction instead of #asTrueFraction in > Float>>adaptToFraction:andCompare: ? > No, it would be a bad thing. For example, it would break the fact that = should be an equivalence relation (a=b && (b=c) ==> (a=c)). See for example http://bugs.squeak.org/view.php?id=3374 for a genesis of this feature. And note that lispers have preceded us by many years. > 0.2 asApproximateFraction ==> (1/5) > 0.2 asTrueFraction ==> (3602879701896397/18014398509481984) > > > There is also asMinimalDecimalFraction or something like that. > > On Jul 11, 2014, at 4:40 PM, Norbert Hartl <norbert(a)hartl.name> wrote: > > > Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: > > > On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: > > On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: > > why not round it before comparing it ? looks like an easy enough problem > to fix . I would expect that to be equal. > > > because it wouldnât be correct :) > > > But then why 4/2 = 2? > > > The same reason why > > 4/3 = 8/6 > > :) > > Norbert > > > > > > On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> > wrote: > >> Thatâs why mathematics and programming are two different thingsâ¦ >> >> >> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> >> wrote: >> >> > Usually comparing against floats is not as deterministic as you would >> > expect. And in other dialects you use #equals: instead of #= >> > >> > But in Pharo there is no #equals: and instead there is a #closeTo:, >> > but #closeTo: has an arbitrary decimal precision, more than enough for >> > commong arithmetic. >> > >> > Regards! >> > >> > Esteban A. Maringolo >> > >> > >> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >> >> :) >> >> >> >> in Smalltalk, the division of two integers is a fraction, not a float. >> >> >> >> i.e. 1/5 is 1/5. >> >> >> >> Thierry >> >> >> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >> >>> >> >>> Hello. >> >>> I found interesting thing: >> >>> Why it is like this? >> >>> >> >>> Best regards, >> >>> Natalia >> >>> >> >> >> >> -- >> >> Thierry Goubier >> >> CEA list >> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >> >> 91191 Gif sur Yvette Cedex >> >> France >> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >> >> >> > >> >> >> > > > > > >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Natalia Tymchuk

Thanks all of you. Best regards, Natalia On Jul 11, 2014, at 5:14 PM, Camille Teruel <camille.teruel(a)gmail.com> wrote: > > On 11 juil. 2014, at 17:02, Natalia Tymchuk <natalia.tymchuk(a)unikernel.net> wrote: > >> But why >> 185/10 = 18.5 -> true? > > Because 18.5 has an exact float representation, while 0.2 hasn't. > > Couldn't we use #asApproximateFraction instead of #asTrueFraction in Float>>adaptToFraction:andCompare: ? > 0.2 asApproximateFraction ==> (1/5) > 0.2 asTrueFraction ==> (3602879701896397/18014398509481984) > > >> >> On Jul 11, 2014, at 4:40 PM, Norbert Hartl <norbert(a)hartl.name> wrote: >> >>> >>> Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: >>> >>>> >>>> On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: >>>> >>>>> On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: >>>>> >>>>>> why not round it before comparing it ? looks like an easy enough problem to fix . I would expect that to be equal. >>>>> >>>>> because it wouldnât be correct :) >>>> >>>> But then why 4/2 = 2? >>> >>> The same reason why >>> >>> 4/3 = 8/6 >>> >>> :) >>> >>> Norbert >>>> >>>>> >>>>>> >>>>>> >>>>>> On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> wrote: >>>>>> Thatâs why mathematics and programming are two different thingsâ¦ >>>>>> >>>>>> >>>>>> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: >>>>>> >>>>>> > Usually comparing against floats is not as deterministic as you would >>>>>> > expect. And in other dialects you use #equals: instead of #= >>>>>> > >>>>>> > But in Pharo there is no #equals: and instead there is a #closeTo:, >>>>>> > but #closeTo: has an arbitrary decimal precision, more than enough for >>>>>> > commong arithmetic. >>>>>> > >>>>>> > Regards! >>>>>> > >>>>>> > Esteban A. Maringolo >>>>>> > >>>>>> > >>>>>> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >>>>>> >> :) >>>>>> >> >>>>>> >> in Smalltalk, the division of two integers is a fraction, not a float. >>>>>> >> >>>>>> >> i.e. 1/5 is 1/5. >>>>>> >> >>>>>> >> Thierry >>>>>> >> >>>>>> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >>>>>> >>> >>>>>> >>> Hello. >>>>>> >>> I found interesting thing: >>>>>> >>> Why it is like this? >>>>>> >>> >>>>>> >>> Best regards, >>>>>> >>> Natalia >>>>>> >>> >>>>>> >> >>>>>> >> -- >>>>>> >> Thierry Goubier >>>>>> >> CEA list >>>>>> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >>>>>> >> 91191 Gif sur Yvette Cedex >>>>>> >> France >>>>>> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >>>>>> >> >>>>>> > >>>>>> >>>>>> >>>>>> >>>>> >>>> >>> >> >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Camille Teruel

On 11 juil. 2014, at 17:02, Natalia Tymchuk <natalia.tymchuk(a)unikernel.net> wrote: > But why > 185/10 = 18.5 -> true? Because 18.5 has an exact float representation, while 0.2 hasn't. Couldn't we use #asApproximateFraction instead of #asTrueFraction in Float>>adaptToFraction:andCompare: ? 0.2 asApproximateFraction ==> (1/5) 0.2 asTrueFraction ==> (3602879701896397/18014398509481984) > > On Jul 11, 2014, at 4:40 PM, Norbert Hartl <norbert(a)hartl.name> wrote: > >> >> Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: >> >>> >>> On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: >>> >>>> On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: >>>> >>>>> why not round it before comparing it ? looks like an easy enough problem to fix . I would expect that to be equal. >>>> >>>> because it wouldnât be correct :) >>> >>> But then why 4/2 = 2? >> >> The same reason why >> >> 4/3 = 8/6 >> >> :) >> >> Norbert >>> >>>> >>>>> >>>>> >>>>> On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> wrote: >>>>> Thatâs why mathematics and programming are two different thingsâ¦ >>>>> >>>>> >>>>> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: >>>>> >>>>> > Usually comparing against floats is not as deterministic as you would >>>>> > expect. And in other dialects you use #equals: instead of #= >>>>> > >>>>> > But in Pharo there is no #equals: and instead there is a #closeTo:, >>>>> > but #closeTo: has an arbitrary decimal precision, more than enough for >>>>> > commong arithmetic. >>>>> > >>>>> > Regards! >>>>> > >>>>> > Esteban A. Maringolo >>>>> > >>>>> > >>>>> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >>>>> >> :) >>>>> >> >>>>> >> in Smalltalk, the division of two integers is a fraction, not a float. >>>>> >> >>>>> >> i.e. 1/5 is 1/5. >>>>> >> >>>>> >> Thierry >>>>> >> >>>>> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >>>>> >>> >>>>> >>> Hello. >>>>> >>> I found interesting thing: >>>>> >>> Why it is like this? >>>>> >>> >>>>> >>> Best regards, >>>>> >>> Natalia >>>>> >>> >>>>> >> >>>>> >> -- >>>>> >> Thierry Goubier >>>>> >> CEA list >>>>> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >>>>> >> 91191 Gif sur Yvette Cedex >>>>> >> France >>>>> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >>>>> >> >>>>> > >>>>> >>>>> >>>>> >>>> >>> >> >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Luc Fabresse

Hi Natalia, The major problem comes from float memory representation. You can read the beginning of chapter "Fun with Floats" of Deep into Pharo that gives examples of why comparing floats is dangerous. http://rmod.lille.inria.fr/deepIntoPharo/ #Luc 2014-07-11 17:02 GMT+02:00 Natalia Tymchuk <natalia.tymchuk(a)unikernel.net>: > But why > 185/10 = 18.5 -> true? > > On Jul 11, 2014, at 4:40 PM, Norbert Hartl <norbert(a)hartl.name> wrote: > > > Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: > > > On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: > > On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: > > why not round it before comparing it ? looks like an easy enough problem > to fix . I would expect that to be equal. > > > because it wouldnât be correct :) > > > But then why 4/2 = 2? > > > The same reason why > > 4/3 = 8/6 > > :) > > Norbert > > > > > > On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> > wrote: > >> Thatâs why mathematics and programming are two different thingsâ¦ >> >> >> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> >> wrote: >> >> > Usually comparing against floats is not as deterministic as you would >> > expect. And in other dialects you use #equals: instead of #= >> > >> > But in Pharo there is no #equals: and instead there is a #closeTo:, >> > but #closeTo: has an arbitrary decimal precision, more than enough for >> > commong arithmetic. >> > >> > Regards! >> > >> > Esteban A. Maringolo >> > >> > >> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >> >> :) >> >> >> >> in Smalltalk, the division of two integers is a fraction, not a float. >> >> >> >> i.e. 1/5 is 1/5. >> >> >> >> Thierry >> >> >> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >> >>> >> >>> Hello. >> >>> I found interesting thing: >> >>> Why it is like this? >> >>> >> >>> Best regards, >> >>> Natalia >> >>> >> >> >> >> -- >> >> Thierry Goubier >> >> CEA list >> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >> >> 91191 Gif sur Yvette Cedex >> >> France >> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >> >> >> > >> >> >> > > > > >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Camille Teruel

On 11 juil. 2014, at 17:01, Igor Stasenko <siguctua(a)gmail.com> wrote: > I would put it this way: > - If you writing code, which requires comparing two floating-point values for equality, then you do something wrong. :D Rule of thumb: don't use floats ;) > > -- > Best regards, > Igor Stasenko.

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Esteban A. Maringolo

2014-07-11 12:02 GMT-03:00 Natalia Tymchuk <natalia.tymchuk(a)unikernel.net>: > But why > 185/10 = 18.5 -> true? It is not deterministic. That's why you can't rely on such comparisons. Regards!

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Natalia Tymchuk

But why 185/10 = 18.5 -> true? On Jul 11, 2014, at 4:40 PM, Norbert Hartl <norbert(a)hartl.name> wrote: > > Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: > >> >> On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: >> >>> On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: >>> >>>> why not round it before comparing it ? looks like an easy enough problem to fix . I would expect that to be equal. >>> >>> because it wouldnât be correct :) >> >> But then why 4/2 = 2? > > The same reason why > > 4/3 = 8/6 > > :) > > Norbert >> >>> >>>> >>>> >>>> On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> wrote: >>>> Thatâs why mathematics and programming are two different thingsâ¦ >>>> >>>> >>>> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: >>>> >>>> > Usually comparing against floats is not as deterministic as you would >>>> > expect. And in other dialects you use #equals: instead of #= >>>> > >>>> > But in Pharo there is no #equals: and instead there is a #closeTo:, >>>> > but #closeTo: has an arbitrary decimal precision, more than enough for >>>> > commong arithmetic. >>>> > >>>> > Regards! >>>> > >>>> > Esteban A. Maringolo >>>> > >>>> > >>>> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >>>> >> :) >>>> >> >>>> >> in Smalltalk, the division of two integers is a fraction, not a float. >>>> >> >>>> >> i.e. 1/5 is 1/5. >>>> >> >>>> >> Thierry >>>> >> >>>> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >>>> >>> >>>> >>> Hello. >>>> >>> I found interesting thing: >>>> >>> Why it is like this? >>>> >>> >>>> >>> Best regards, >>>> >>> Natalia >>>> >>> >>>> >> >>>> >> -- >>>> >> Thierry Goubier >>>> >> CEA list >>>> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >>>> >> 91191 Gif sur Yvette Cedex >>>> >> France >>>> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >>>> >> >>>> > >>>> >>>> >>>> >>> >> >

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Igor Stasenko

I would put it this way: - If you writing code, which requires comparing two floating-point values for equality, then you do something wrong. -- Best regards, Igor Stasenko.

July 11, 2014

Re: [Pharo-dev] Division isn't correct

by Norbert Hartl

Am 11.07.2014 um 16:33 schrieb Yuriy Tymchuk <yuriy.tymchuk(a)me.com>: > > On 11 Jul 2014, at 16:30, Esteban Lorenzano <estebanlm(a)gmail.com> wrote: > >> On 11 Jul 2014, at 16:22, kilon alios <kilon.alios(a)gmail.com> wrote: >> >>> why not round it before comparing it ? looks like an easy enough problem to fix . I would expect that to be equal. >> >> because it wouldnât be correct :) > > But then why 4/2 = 2? The same reason why 4/3 = 8/6 :) Norbert > >> >>> >>> >>> On Fri, Jul 11, 2014 at 5:18 PM, Yuriy Tymchuk <yuriy.tymchuk(a)me.com> wrote: >>> Thatâs why mathematics and programming are two different thingsâ¦ >>> >>> >>> On 11 Jul 2014, at 16:08, Esteban A. Maringolo <emaringolo(a)gmail.com> wrote: >>> >>> > Usually comparing against floats is not as deterministic as you would >>> > expect. And in other dialects you use #equals: instead of #= >>> > >>> > But in Pharo there is no #equals: and instead there is a #closeTo:, >>> > but #closeTo: has an arbitrary decimal precision, more than enough for >>> > commong arithmetic. >>> > >>> > Regards! >>> > >>> > Esteban A. Maringolo >>> > >>> > >>> > 2014-07-11 10:59 GMT-03:00 Goubier Thierry <thierry.goubier(a)cea.fr>: >>> >> :) >>> >> >>> >> in Smalltalk, the division of two integers is a fraction, not a float. >>> >> >>> >> i.e. 1/5 is 1/5. >>> >> >>> >> Thierry >>> >> >>> >> Le 11/07/2014 15:53, Natalia Tymchuk a Ã©crit : >>> >>> >>> >>> Hello. >>> >>> I found interesting thing: >>> >>> Why it is like this? >>> >>> >>> >>> Best regards, >>> >>> Natalia >>> >>> >>> >> >>> >> -- >>> >> Thierry Goubier >>> >> CEA list >>> >> Laboratoire des Fondations des SystÃ¨mes Temps RÃ©el EmbarquÃ©s >>> >> 91191 Gif sur Yvette Cedex >>> >> France >>> >> Phone/Fax: +33 (0) 1 69 08 32 92 / 83 95 >>> >> >>> > >>> >>> >>> >> >

July 11, 2014