| 1 : |
blume |
780 |
<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML//EN">
|
| 2 : |
|
|
<html>
|
| 3 : |
|
|
<head>
|
| 4 : |
|
|
<title>SML/NJ Version 110.32 NEWS</title>
|
| 5 : |
|
|
</head>
|
| 6 : |
|
|
|
| 7 : |
|
|
<body bgcolor="white">
|
| 8 : |
|
|
<center><h1>Standard ML of New Jersey<BR>
|
| 9 : |
|
|
Version 110.32, December 22, 2000</h1>
|
| 10 : |
|
|
</center>
|
| 11 : |
|
|
|
| 12 : |
|
|
<center>
|
| 13 : |
|
|
<a href=http://cm.bell-labs.com/cm/cs/what/smlnj/index.html>
|
| 14 : |
|
|
Home page</a>
|
| 15 : |
|
|
</center>
|
| 16 : |
|
|
|
| 17 : |
|
|
<blockquote>
|
| 18 : |
|
|
<center>
|
| 19 : |
|
|
<h2> Warning </h2>
|
| 20 : |
|
|
</center>
|
| 21 : |
|
|
<blockquote>
|
| 22 : |
|
|
<blockquote>
|
| 23 : |
|
|
<em>
|
| 24 : |
|
|
This version is intended for compiler hackers. The
|
| 25 : |
|
|
version ought to be quite stable, however we have not run
|
| 26 : |
|
|
our full regression testing.
|
| 27 : |
|
|
</em>
|
| 28 : |
|
|
</blockquote>
|
| 29 : |
|
|
</blockquote>
|
| 30 : |
|
|
|
| 31 : |
|
|
|
| 32 : |
|
|
<hr>
|
| 33 : |
|
|
|
| 34 : |
|
|
<center><h2>Summary:</h2></center>
|
| 35 : |
|
|
This version is being used to check point a large number of changes
|
| 36 : |
|
|
to MLRISC, which now uses infinite precision integers on all
|
| 37 : |
|
|
architectures. We have run the Basis regressions tests so the
|
| 38 : |
|
|
changes ought to be fairly robust.
|
| 39 : |
|
|
<p>
|
| 40 : |
|
|
There are a number of other CM, Basis, and Tool related changes as
|
| 41 : |
|
|
well.
|
| 42 : |
|
|
|
| 43 : |
|
|
<hr>
|
| 44 : |
|
|
<center><h3>MLRISC </h3></center>
|
| 45 : |
|
|
<h3> Fixes </h3>
|
| 46 : |
|
|
This change fixes two problems:
|
| 47 : |
|
|
<p>
|
| 48 : |
|
|
|
| 49 : |
|
|
<ol>
|
| 50 : |
|
|
|
| 51 : |
|
|
<p><li> There was a proliferation of constructors in MLRISC to
|
| 52 : |
|
|
handle integers of various sizes, e.g.
|
| 53 : |
|
|
<pre>
|
| 54 : |
|
|
LI of int | LI32 of Int32.int | LI64 of Int64.int
|
| 55 : |
|
|
...
|
| 56 : |
|
|
</pre>
|
| 57 : |
|
|
which are all replaced with
|
| 58 : |
|
|
<pre>
|
| 59 : |
|
|
LI of IntInf.int
|
| 60 : |
|
|
</pre>
|
| 61 : |
|
|
|
| 62 : |
|
|
<p><li> There is the ever nagging question of cross compiling to
|
| 63 : |
|
|
64bits.
|
| 64 : |
|
|
</ol>
|
| 65 : |
|
|
|
| 66 : |
|
|
<h3>Compilation Speed</h3>
|
| 67 : |
|
|
The x86 compiling the compiler is 9% slower than 110.31, however,
|
| 68 : |
|
|
a lot of this, and more, can be recovered. The compiler presently
|
| 69 : |
|
|
uses infinite precision in the front end (AST creation), which is converted to 31
|
| 70 : |
|
|
and 32 bit integers in the middle end (FLINT and CPS), and then
|
| 71 : |
|
|
converted back to infinite precision in MLRISC (with the current
|
| 72 : |
|
|
changes).
|
| 73 : |
|
|
<p>
|
| 74 : |
|
|
Infinite precision in the middle end will allow a uniform
|
| 75 : |
|
|
treatment of optimizations across words and integers of different
|
| 76 : |
|
|
widths/sizes. Currently, most optimizations are biased towards
|
| 77 : |
|
|
integers and words represented in 31 bits only.
|
| 78 : |
|
|
<p>
|
| 79 : |
|
|
|
| 80 : |
|
|
<h3> What's next</h3>
|
| 81 : |
|
|
Not in chronological order:
|
| 82 : |
|
|
|
| 83 : |
|
|
<ol>
|
| 84 : |
|
|
<li> Use infinite precision in the middle end to allow a
|
| 85 : |
|
|
cleaner and more uniform optimization module for arithmetic.
|
| 86 : |
|
|
|
| 87 : |
|
|
<p><li> Support for the generation of infinite precision values
|
| 88 : |
|
|
in the back end.
|
| 89 : |
|
|
|
| 90 : |
|
|
<p><li> Changes to support the single toplevel declaration:
|
| 91 : |
|
|
<pre>
|
| 92 : |
|
|
structure LargeInt = IntInf
|
| 93 : |
|
|
</pre>
|
| 94 : |
|
|
</ol>
|
| 95 : |
|
|
|
| 96 : |
|
|
<hr>
|
| 97 : |
|
|
|
| 98 : |
|
|
<center><h3>Basis </h3></center>
|
| 99 : |
|
|
Added the missing bit level functions to the Basis <tt>IntInf</tt>
|
| 100 : |
|
|
structure.
|
| 101 : |
|
|
|
| 102 : |
|
|
<hr>
|
| 103 : |
|
|
<center><h3>Tools </h3></center>
|
| 104 : |
|
|
Improvements to the 'nowhere' tool to handle OR-patterns.
|
| 105 : |
|
|
The nowhere tool is a syntactic extension that allows for
|
| 106 : |
|
|
guarded patterns. See MLRISC/Tools/Doc/nowhere.tex.
|
| 107 : |
|
|
|
| 108 : |
|
|
<hr>
|
| 109 : |
|
|
<center><h3>CM </h3></center>
|
| 110 : |
|
|
<ul>
|
| 111 : |
|
|
<li> Added a new "dir" tool to CM's toolset; this can simplify writing
|
| 112 : |
|
|
CM descriptions (see the manual)
|
| 113 : |
|
|
|
| 114 : |
|
|
<p><li> Made some modifications to CM's "Tools" API and documented everything
|
| 115 : |
|
|
|
| 116 : |
|
|
<p><li> CM now leaves exceptions (from the compiler) alone if they are not
|
| 117 : |
|
|
expected; this causes such exceptions to travel up to the top-level
|
| 118 : |
|
|
where they are reported as usual
|
| 119 : |
|
|
|
| 120 : |
|
|
<p><li> Eliminated the "[wrote ...]" progress message in favor of reporting
|
| 121 : |
|
|
some size information for the generated binfile instead
|
| 122 : |
|
|
|
| 123 : |
|
|
<p><li> Removed source code for some project-in-progress that is not ready
|
| 124 : |
|
|
for prime-time (yet)
|
| 125 : |
|
|
|
| 126 : |
|
|
<p><li> Made anchor environments ("bind:"-option for class "cm") inter-operate
|
| 127 : |
|
|
in a more useful way with CMB.make
|
| 128 : |
|
|
</ol>
|
| 129 : |
|
|
<hr>
|
| 130 : |
|
|
<font size=-2>
|
| 131 : |
|
|
<address><a href="mailto:george@research.bell-labs.com">
|
| 132 : |
|
|
Lal George</a></address>
|
| 133 : |
|
|
<!-- Created: Thu Aug 6 00:13:09 EDT 1998 -->
|
| 134 : |
|
|
<!-- hhmts start -->
|
| 135 : |
|
|
Last modified: Sat Dec 23 11:23:17 EST 2000
|
| 136 : |
|
|
<!-- hhmts end -->
|
| 137 : |
|
|
</font>
|
| 138 : |
|
|
</blockquote>
|
| 139 : |
|
|
</body>
|
| 140 : |
|
|
</html>
|
Parent Directory
| 
Revision Log
