1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
|
<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN"
"http://www.w3.org/TR/html4/loose.dtd">
<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=iso-8859-1">
<link rel="stylesheet" type="text/css" href="../../../../boost.css">
<title>Interval-TODO.htm</title>
</head>
<body lang="en">
<h1>TODO list for the Interval Arithmetic library</h1>
<h2>Comments from the review process</h2>
<ul>
<li>It would be nice to have a 100% portable Rouding policy class based on
LIA-1 only, with no additional requirement such as IEEE 754 or even
more.</li>
<li>For pi and other constants, we should fetch them from the Math
constants library when it is ready.</li>
</ul>
<h2>Various items</h2>
<ul>
<li>Support the <code>-mfpmath=sse</code> option of g++ for Pentium IV
optimized code. This requires a different rounding policy class.</li>
<li>Have an optimized specialization of <code>interval<double></code>
using SSE2 instructions? Tricky.</li>
<li>Write a traits specialization for <code>interval<MPFR></code>,
and other non-built-in number types.</li>
<li>Have a robust implementation of transcendental functions over the
built-in floating point types (e.g. using MPFR)?</li>
<li>Safe conversions from <code>interval<double></code> to
<code>interval<float></code>? Requires partial specialization.</li>
<li>It would be nice to use the expression template mecanism to
automatically use the more efficient unprotected rounding mode version
for small subexpressions (although you can do bigger expressions by
hand).</li>
</ul>
<hr>
<p>Revised: 2003-01-19<br>
Copyright (c) Guillaume Melquiond, Sylvain Pion, Herv Brnnimann, 2002.<br>
Polytechnic University.</p>
</body>
</html>
|
