racket/doc/srfi-std/srfi-7.html
Matthew Flatt 28a3f3f0e7 r5rs and srfi docs and bindings
svn: r9336
2008-04-16 20:52:39 +00:00

338 lines
11 KiB
HTML

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 3.2//EN">
<html>
<head>
<title>SRFI 7: Feature-based program configuration language</title>
</head>
<body>
<H1>Title</H1>
SRFI-7: Feature-based program configuration language
<H1>Author</H1>
Richard Kelsey
<H1>Status</H1>
This SRFI is currently in ``final'' status. To see an explanation of each status that a SRFI can hold, see <A HREF="http://srfi.schemers.org/srfi-process.html">here</A>.
You can access the discussion via <A HREF="http://srfi.schemers.org/srfi-7/mail-archive/maillist.html">the archive of the mailing list</A>.
<P><UL>
<LI>Received: 1999/05/12
<LI>Draft: 1999/05/26-1999/07/26
<LI>Final: 1999/08/19
</UL>
<H1>Abstract</H1>
<p>
This SRFI describes a configuration language to be used for specifing
the set of Scheme features or extensions required to run a program.
In addition to a list of required features, a program may also contain
Scheme code to be used only when a particular feature or combination of
features is available.
</p>
<p>
The configuration language is entirely distinct from Scheme; it is
neither embedded in Scheme nor includes Scheme as a subset.
</p>
<H1>Rationale</H1>
<p>
The use of a separate configuration language makes it easy for
both human and machine readers to determine the requirements of a
program. It also avoids ambiguities arising from having to expand
macros to determine which features, and thus which macros, are used
in the program.
</p>
<p> See <a href="http://srfi.schemers.org/srfi-0">SRFI 0</a> for a
rationale for the need for some kind of configuration control.
</p>
<H1>Specification</H1>
<a name="program"></a>
<a name="requires"></a>
<a name="code"></a>
<a name="feature-cond"></a>
<a name="and"></a>
<a name="or"></a>
<a name="not"></a>
<a name="else"></a>
<H2>Syntax</H2>
<p>
<pre>
&lt;program&gt; --&gt; (program &lt;program clause&gt;+)
&lt;program clause&gt;
--&gt; (requires &lt;feature identifier&gt;+)
| (files &lt;filename&gt;*)
| (code &lt;Scheme expression, definition, or syntax definition&gt;*)
| (feature-cond &lt;feature-cond clause&gt;+)
| (feature-cond &lt;feature-cond clause&gt;* (else &lt;program clause&gt;+))
&lt;feature-cond clause&gt;
--&gt; (&lt;feature requirement&gt; &lt;program clause&gt;+)
&lt;feature requirement&gt;
--&gt; &lt;feature identifier&gt;
| (and &lt;feature requirement&gt;*)
| (or &lt;feature requirement&gt;*)
| (not &lt;feature requirement&gt;)
&lt;feature identifier&gt;
--&gt; a symbol which is the name of a SRFI
</pre>
</p>
<H2>Semantics</H2>
<p>
The configuration language is distinct from Scheme. Given a set of
available features a <code>&lt;program&gt;</code> can be converted into a
sequence of Scheme commands, definitions, and syntax definitions.
This conversion does not require expanding macros or doing any other
processing of Scheme code.
</p>
<p>
An implementation of the configuration language will need to provide
some method for executing a program. For example, it might have a
<code>(LOAD-PROGRAM &lt;filename&gt;)</code> function or a compiler that
compiles a program into an executable file.
</p>
<p> The meanings of the different <code>&lt;program&gt;</code> clauses are
given below. The ordering of the clauses in a <code>&lt;program&gt;</code>
determines the order of the forms in the resultant Scheme program.
<p> In processing the <code>REQUIRES</code> and
<code>FEATURE-COND</code> clauses in a <code>&lt;program&gt;</code>,
an implementation should be consistent with some fixed set of present
and absent features. An implementation may analyze a <code>&lt;program&gt;</code>
before choosing a set of features to use in processing it, but it
should not use different sets of features for different clauses in the
same <code>&lt;program&gt;</code>.
</p>
<dl>
<dt><code>(requires &lt;feature-identifier&gt;+)</code></dt>
<dd>The listed features are required by the program. If one or
more is not provided by the implementation the program cannot
be run.
</dd>
<dt><code>(files &lt;filename&gt;*)</code></dt>
<dd>The listed files are read and the forms they contain added to the program.</dd>
<dt><code>(code &lt;body&gt;)</code></dt>
<dd>The forms in &lt;body&gt; are added to the program.</dd>
<dt><code>(feature-cond &lt;feature-cond clause&gt;+)</code></dt>
<dd>The meaning of a <code>FEATURE-COND</code> clause is that of the
<code>&lt;program-clause&gt;</code>s in the first <code>&lt;feature-cond clause&gt;</code> whose
<code>&lt;implementation-requirement&gt;</code> is satisfied by the implementation.
If an <code>ELSE</code> clause is present it is used if and only if no preceding
clause is satisfied; a <code>FEATURE-COND</code> with an
<code>ELSE</code> clause is always satisfied.
<p>If no clause can be satisified the <code>&lt;program&gt;</code> cannot be evaluated in
the implementation.</p>
<p>
The meaning of the <code>&lt;implementation requirement&gt;</code>s is as follows:</p>
<table>
<tr><td><code>&lt;feature identifier&gt;</code></td><td>satisfied if the feature is present</td>
</tr>
<tr><td><code>(and)</code></td><td>always satisfied</td>
</tr>
<tr><td><code>(and x ...)</code></td><td>satisfied if every
<code>X</code> is satisfied</td>
</tr>
<tr><td><code>(or)</code></td><td>never satisfied</td></tr>
<tr><td><code>(or x ...)</code></td><td>satisfied if any
<code>X</code> is satisfied</td></tr>
<tr><td><code>(not x)</code></td><td>satisfied if <code>X</code> is not satisfied</td></tr>
</table>
</dl>
<H1>Implementation</H1>
<p> Two implementations are provided here. The first is a
<code>PROCESS-PROGRAM</code> function that converts a
<code>&lt;program&gt;</code>, represented as an S-expression, and a
list of feature identifiers and returns the list expressions,
definitions, and syntax definitions that are the source for the
<code>&lt;program&gt;</code> in the presence of those features. The
function returns <code>#F</code> if the program cannot be run using
the features provided.
</p>
<p>
This is not a complete implementation of the configuration language; it needs
an (implementation-dependent) method for evaluating the forms returned by
<code>PROCESS-PROGRAM</code>.
</p>
<p>
<pre>
(define (process-program program features)
(call-with-current-continuation
(lambda (exit) ; We exit early when an unsatisfiable clause is found.
; Process each clause in turn
(define (process-clauses clauses)
(if (null? clauses)
'()
(append (process-clause (car clauses))
(process-clauses (cdr clauses)))))
; Dispatch on the type of the clause.
(define (process-clause clause)
(case (car clause)
((requires)
(if (all-satisfied? (cdr clause))
'()
(exit #f)))
((code)
(cdr clause))
((files)
(read-files (cdr clause)))
((feature-cond)
(process-cond-clauses (cdr clause)))))
; Loop through CLAUSES finding the first that is satisfied.
(define (process-cond-clauses clauses)
(cond ((null? clauses)
(exit #f))
((or (and (eq? (caar clauses) 'else)
(null? (cdr clauses)))
(satisfied? (caar clauses)))
(process-clauses (cdar clauses)))
(else
(process-cond-clauses (cdr clauses)))))
; Compound requirements are handled recursively, simple ones are tested.
(define (satisfied? requirement)
(if (pair? requirement)
(case (car requirement)
((and)
(all-satisfied? (cdr requirement)))
((or)
(any-satisfied? (cdr requirement)))
((not)
(not (satisfied? (cadr requirement)))))
(memq requirement features)))
; True if every requirement in LIST is satisfied.
(define (all-satisfied? list)
(if (null? list)
#t
(and (satisfied? (car list))
(all-satisfied? (cdr list)))))
; True if any requirement in LIST is satisfied.
(define (any-satisfied? list)
(if (null? list)
#f
(or (satisfied? (car list))
(any-satisfied? (cdr list)))))
; Start by doing the whole program.
(process-clauses (cdr program)))))
; Returns a list of the forms in the named files.
(define (read-files filenames)
(if (null? filenames)
'()
(append (call-with-input-file (car filenames)
(lambda (in)
(let label ()
(let ((next (read in)))
(if (eof-object? next)
'()
(cons next (label)))))))
(read-files (cdr filenames)))))
</pre></p>
<p>
The second implementation is a <code>PROGRAM</code> macro that implements
the configuration language in terms of the <code>COND-EXPAND</code>
syntax of <a href="http://srfi.schemers.org/srfi-0">SRFI 0</a>.
Note that this implementation requires that <code>LOAD</code> use the current
evaluation environment.
</p>
<p><pre>
(define-syntax program
(syntax-rules (requires files code feature-cond)
((program)
(begin))
((program (requires feature-id ...)
more ...)
(begin (cond-expand ((and feature-id ...) 'okay))
(program more ...)))
((program (files filename ...)
more ...)
(begin (load filename) ...
(program more ...)))
((program (code stuff ...)
more ...)
(begin stuff ...
(program more ...)))
((program (feature-cond (requirement stuff ...) ...)
more ...)
(begin (cond-expand (requirement (program stuff ...)) ...)
(program more ...)))))
</pre></p>
<H1>Copyright</H1>
Copyright (C) Richard Kelsey (1999). All Rights Reserved.
<p>
Permission is hereby granted, free of charge, to any person obtaining
a copy of this software and associated documentation files (the
"Software"), to deal in the Software without restriction, including
without limitation the rights to use, copy, modify, merge, publish,
distribute, sublicense, and/or sell copies of the Software, and to
permit persons to whom the Software is furnished to do so, subject to
the following conditions:
</p>
<p>
The above copyright notice and this permission notice shall be
included in all copies or substantial portions of the Software.
</p>
<p>
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND,
EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF
MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE
LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION
OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION
WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
</p>
<hr>
<address>Editor: <a href="mailto:srfi-editors@srfi.schemers.org">Mike Sperber</a></address>
</body>
</html>