mathematical problem with Int()

[jennico]

well, i think my presumption to implement this automatic rounding option could be worth thinking about.

in my script i was using the decimal places like bits: 8.0010010110 in order to save additional data to an integer. the simple integer (8) determines the image to be displayed and the decimal places mark certain individual properties of the image (1 has the specific property, 0 does not). this method seemed very clean to me, because i can save lots of data without defining arrays.

but then it happened what i could not expect: for some reasons the data was false after adding and subtracting 1s and 0s and changing the integer value and i had to look for the error quite a long time (three days). i was really surprised about this floating point problem when i hit it.

so, maybe someone should think about this and maybe there will be added such an option, if someone like me needs "clean math". in this case, when the machine cannot cope with it, it could at least be optionally forced to pretend so.

i agree to your goto opinion in general, in 99% of the cases you can use (though sometimes huge) loops instead. but there are still left some rare cases in which you could use goto very well, e.g. when you want to restart a programm from the very first line. but there is no basic problem in living without goto.

so long j. :-)

btw: i studied physics and math in university and when you want to fly to the moon i tell you better don't leave the results unrounded... :-) to sum it up: there is a difference between math and informatics you find the same differences when you compare math with physics.

Edited May 10, 2007 by jennico

[PsaltyDS]

in my script i was using the decimal places like bits: 8.0010010110 in order to save additional data to an integer. the simple integer (8) determines the image to be displayed and the decimal places mark certain individual properties of the image (1 has the specific property, 0 does not). this method seemed very clean to me, because i can save lots of data without defining arrays.

You are creating a data word that contains flags, a very common programming device.

All you have to do is define each flag as a power of 2 instead of 10 - and your problems are solved!

; Flags for attributes
$AttribA = 1
$AttribB = 2
$AttribC = 4
$AttribD = 8
$AttribE = 16
$AttribF = 32
$AttribG = 64
$AttribH = 128

; Default = 9 which is attributes A and D with no others:
$CurrentAttribs = $AttribA + $AttribD

; Set Attrib F without changing any others
$CurrentAttribs = BitOR($CurrentAttribs, $AttribF)

; Clear Attrib D without changing any others
$CurrentAttribs = BitAND($CurrentAttribs, BitNOT($AttribD))

This works much better than decimal flags because you easily work with individual flags by way of BitAnd() , BitOr(), etc.

You could set or clear an attribute by addition/subtraction, but it all goes haywire if you are re-setting/re-clearing it, because all the other values get unintentionally modified.

Edited May 10, 2007 by PsaltyDS

[jennico]

hi PsaltyDS,

thank you for that. of course this was my first idea using the dual system in order to add the data to a value, but i decided to use I/O because it's easier to keep an overwiew, and from a different point of view, 8.1 is the same as 8.1, while 8.01 means 8.2, 8.001 8.4, 8.0001 8.8 and so on. and as you say, the other values will not be changed unintentionally when using I/O.

BitAnd(),BitOr() and so on is certainly the right way to calculate binary data. well, now i have made my way with I/O, and i only have to keep in mind better to round the differences (luckily addition is no problem for the artificial brain). though better to me still appears the option to establish a global rule for that (opt()).

and i will be going on stating that 8.01 - 8 results 0.01 or 1x10-² and nothing else in this world as long as you don't lose your common sense and your organic brain.

thanx for your help

[PsaltyDS]

...and from a different point of view, 8.1 is the same as 8.1, while 8.01 means 8.2, 8.001 8.4, 8.0001 8.8 and so on.

You can't have 8 to the left of the decimal place for a binary number. Decimal 8.5 would be 1000.1 in binary (2³ + 2^-1)

Binary = Decimal

0.1 = 1/2 (0.5)

0.01 = 1/4 (0.25)

0.001 = 1/8 (0.125)

0.0001 = 1/16 (0.0625)

0.00001 = 1/32 (0.03125)

etc.

[jennico]

yeah, okay, the topic is all about the decimal places without integer, that's what i was referring to. that's why i need

$decimalplaces_of_value=round($value-int($value),15)

but now it gets confusing i think.

j.

[The Kandie Man]

"Incoming!"

*Boom*

"Return fire!"

*Boom*

- The Kandie Man

[martin]

well, i know that i can round the result. but still the result before rounding is wrong. i used this operation in a program and couldn't find this error in several days.

8.0001 - 8 should return 0.0001 without rounding, or not ?

every cheap calculator does it right. (btw: the windows calculator does it, too)

j. :-)

You're kidding us?

You needed to find an error of about 1 part in 10^14 and took several days on it?

Have they got any spare jobs at your place?

Edited May 12, 2007 by martin

[PsaltyDS]

You're kidding us?

You needed to find an error of about 1 part in 10^14 and took several days on it?

Have they got any spare jobs at your place?

I haven't minded the conversation at all. Had to dredge up multi-base math that I haven't thought about since DS 'A' school over 25 years ago.