_Float

 A module to manipulate double-precision floating point numbers.

]
[Introduction

]
[Using Float

 The Float module is loaded most easily by running the !Float application on boot-up. The method used will depend on the version of !Boot that is installed.
]
[Copyright

 Float and its documentation are  Steve Drain, Kappa.

. mailto:steve@kappa.me.uk
. http://steve@kappa.me.uk
]
[Appendix 1 - Commands


}
]
[Appendix 2 - SWIs


}
]
[Appendix 3 - History

{Version 0.30 (18 Dec 2013)

 Now have a copy of "Computer Approximations", so can start to implement coefficients matched to required precision and explore exponential functions.

 Trig functions adapted with new coefficients have reduced number of terms and are a little faster.
}
{Version 0.26 (16 Dec 2013)

 Realised that slowness is due to the context switching in each SWI.

 Module re-written so that calling program is responsible for context switching.Retain a context check to avoid calling VFP code without a context.
}
{Version 0.24 (13 Dec 2013)

 All trig functions re-written.

 Speed tests show only a small, 30%, advantage over FPE. This make the project seem rather impractical.
}
{Version 0.20 (08 Dec 2013)

 Discovered paper showing how to implement higher precison.

 Cos re-written and now matches or exceeds FPE.
}
{Version 0.18 (06 Dec 2013)

 All trig functions written using coefficients from web, but precision is not comparable to FPE.
}
{Version 0.16 (28 Nov 2013)

 Val SWI, to read a value from a string, completed for FPE and VFP. Method adapted from BASIC module.
}
{Version 0.11 (25 Nov 2013)

 VFP context is now set up within the module if possible, and switched for each SWI as required.

 Hiccup over endian status resolved!
}
{Version 0.10 (11 Nov 2013)

 Original experimental version.
}
]
[Appendix 3 - Feedback

 I would welcome comments on any aspect of Float, as well as the documentation.
]
