Contributor: DIMITRI SMITS (* This unit has been typed from the top of my head, so test it first with the examples in my previous message to Eli. As a matter of fact, this is a Unit that is based upon my explanations to Eli, so it somewhat NEEDS it ;) But I tested it though, just before posting it ;) This unit may be published in the next SWAGs (at least I think something like this belongs at least in the MISC SWAG becos it doesn't have one yet (I think).) but include the other mail too! (the previous one) If any questions, direct them to Gongo/Insecabilis dsmits@zorro.ruca.ua.ac.be or 2:292/8013.12 (fido) *) Unit FixP; { (c) 1996 by Dimitri Smits aka Gongo/Insecabilis Released to the public Domain on june 22 '96 You may use this in any production you want just notice me of the result, always happy to see such stuff ;) ... not needed to credit me, just greet me (either in the documentation or in the demo/game/softpackage ;) } INTERFACE TYPE fp = LONGINT; { 16.16 fixed point, signed So all numbers -32768.0000 <= x < 32767.9999 should be sufficiently supported (although mul and div might lose some precision, or give odd results for large numbers (integer part)} FUNCTION fp_add ( fp1,fp2: fp): fp; { Not really needed, but when going to overloading functions in C++ or so, this method is needed ;) in other words, just here to make things complete } { an fp3 := fp1 + fp2 is enough :) } FUNCTION fp_sub ( fp1,fp2: fp): fp; { same as with fp_add, only + is - now } FUNCTION fp_mul ( fp1,fp2: fp): fp; FUNCTION fp_div ( fp1,fp2: fp): fp; FUNCTION fp2float (fpt : fp) : REAL; FUNCTION float2fp (fl : REAL) : fp; IMPLEMENTATION FUNCTION fp_add (fp1,fp2 : fp) : fp; BEGIN fp_add := fp1 + fp2; END; FUNCTION fp_sub (fp1,fp2 : fp) : fp; BEGIN fp_sub := fp1 - fp2; END; FUNCTION fp_mul (fp1,fp2 : fp) : fp; BEGIN IF abs(fp1) > abs(fp2) THEN fp_mul := (fp1 SHR 8 * fp2) SHR 8 ELSE fp_mul := (fp2 SHR 8 * fp1) SHR 8; {16-bit precision needed, not 32 =)} END; FUNCTION fp_div (fp1,fp2 :fp) : fp; BEGIN fp_div := (fp1 SHL 8) DIV (fp2 SHR 8); { May lose some precision there} END; FUNCTION fp2Float (fpt : fp) : REAL; BEGIN fp2Float := fpt / 65536; END; FUNCTION float2fp (fl : REAL) : fp; BEGIN float2fp := ROUND (fl * 65536); END; BEGIN END.