SRFIs

Bigloo supports various SRFIs (Scheme Request For Implementation). Some of them are integrated in the Bigloo core libraries. Some others are implemented by the means of Bigloo libraries (see Bigloo Libraries). Only the first ones are described in the manual. Check the Bigloo web page (http://www-sop.inria.fr/indes/fp/Bigloo). The current Bigloo core library support the following SRFIs:

SRFI 0

cond-expand [clause]bigloo syntax

The cond-expand form tests for the existence of features at macro-expansion time. It either expands into the body of one of its clauses or signals and error during syntactic processing. cond-expand expands into the body of the first clause whose feature requirement is currently satisfied (the else clause, if present, is selected if none of the previous clauses is selected).

A feature requirement has an obvious interpretation as a logical formula, where the variables have meaning true is the feature corresponding to the feature identifier, as specified in the SRFI registry, is in effect at the location of the cond-expand form, and false otherwise. A feature requirement is satisfied it its formula is true under this interpretation. The formula may make use of identifier, and, or and not operators.

Since Bigloo version 3.1b, cond-expand formula may use the new library operator that checks if a library exists and is available. Its syntax is: (library <libname>).

Examples: 

(write (cond-expand
          (srfi-0 (* 1 2))
          ((or (library fthread) (library pthread)) (- 4 1))
          (else (+ 3 4))))
   ⇥ 2

(cond-expand 
   (bigloo (define (command-line-arguments) (command-line)))
   (else (define (command-line-arguments) '())))
The second example assumes that bigloo is an alias for the SRFI associated with the specification of Bigloo (i.e. the documentation for that Scheme system).

Since Bigloo 3.4b, cond-expand formula may use the new config operator that checks the value of a configuration entry. Its syntax is: (config endianeness little-endian). This feature relies on the bigloo-config function. See System Programming for additional details.

When writing portable code, the case used for the feature identifier should match the one in the SRFI registry. This is to ensure that the feature identifier will be correctly recognized whether or not the Scheme system is case-sensitive. To support case-insensitive Scheme systems, the feature identifiers in the SRFI registry are guaranteed to be unique even when ignoring the case.

In order to distinguish Bigloo versions, the following symbols are recognized in cond-expand forms.

When finalizers have been configured, the two following symbols are recognized by cond-expand:

Bigloo implements differents SRFI for the compiler and the interpreter. Thus, their are two Bigloo SRFI registers. One for the compiler and one for the interpreter. Bigloo compiler SRFI register contains at least the following symbols:

With respect to the currently used Bigloo back-end, one of these symbols is registered:

Bigloo compiler implements the following SRFI:

Then the -g flag is used, the Bigloo compiler additionally implements the SRFI:

Bigloo interpreter implements the following SRFI:

When a library is used, the name of the library is added to the compiler SRFI register. That is: 

(module foo
   (library srfi1))

(print (cond-expand (srfi1 'with-srfi1) (else 'nothing)))
   ⇥ 'with-srfi1
(print (eval '(cond-expand (srfi1 'with-srfi1) (else 'nothing))))
   ⇥ 'with-srfi1
A property representing actual integers bit size is defined:

The frequently defined values are:

Other values could be observed in the future. Note that the actual values of a particular setting can be obtained with: 

(bigloo-config 'int-size)
(bigloo-config 'elong-size)
A configuration can be tested with:

For instance: 

(cond-expand
  ((and bigloo-c (config have-c99-stack-alloc #t)) ...)
  ...)
.keep

register-srfi! srfi-namebigloo procedure

unregister-srfi! srfi-namebigloo procedure

unregister-eval-srfi! srfi-namebigloo procedure

register-compile-srfi! srfi-namebigloo procedure

unregister-compile-srfi! srfi-namebigloo procedure

This argument srfi-name is a symbol. It registers srfi-name in the Bigloo interpreter SRFI register. This function must only be used when implementing a library. The code of that library must contain one unique call to register-eval-srfi!. Let's suppose, for instance, a format library. The implementation for that library must contain an expression like:

The functions unregister-XXX-srfi! unregisters a srfi. 

(register-eval-srfi! 'format)
Calling (register-eval-srfi! name) makes name supported by interpreted cond-expand forms.

Note: There is no register-compiler-srfi! because the compiler automatically registers SRFI when the -library flags are used. However, it exists several ways to tell the compiler that it actually supports some srfis when compiling some modules.

.keep

SRFI 1

The SRFI 1 is implemented as a Bigloo library. Hence, in order to use the functions it provides, a module must import it. 

(module ex
   (library srfi1))

(print (find-tail even? '(3 1 37 -8 -5 0 0)))
 ⇒ '(-8 -5 0 0))

SRFI 22

The SRFI 22 describes basic prerequisites for running Scheme programs as Unix scripts in a uniform way. A file (henceforth a scipt) conforming SRFI 22 has the following syntax: 

<script>         ⇒ <script prelude>? <program>
<script prelude> ⇒ #! <space> <all but linebreak>* <linebreak>
A Scheme script interpreter loads the <script>. It ignores the script prelude and interprets the rest of the file according to the language dialect specified by the name of the interpreter.

The Scheme script interpreter may also load a different file after making a reasonable check that loading it is semantically equivalent to loading <script>. For example, the script interpreter may assume that a file with a related name (say, with an additional extension) is a compiled version of <script>.

An example of SRFI-22 script

Let us consider the following Bigloo script located in a file `foo.scm': 

#! /usr/bin/env ./execute
(module foo
   (main main))

(define (main argv)
   (print "foo: " argv))
Let us consider the following `execute' shell script: 

$ cat > execute
#!/bin/sh
bigloo -i $*
Provided that `foo.scm' as the execute flag switched on, it is possible to execute it: 

$ chmod u+x foo.scm
$ ./foo.scm
  ⇥ foo: (./foo.scm)
The same Bigloo module can be compiled and executed such as: 

$ bigloo foo.scm
$ ./a.out
  ⇥ foo: (a.out)

Lazy compilation with SRFI-22

SRFI-22 can be used to implement lazy compilation. For instance, let us consider the following shell script: 

$ cat > bgl
#!/bin/sh
SOURCEFILE=$1
case $SOURCEFILE in
    *.scm)
        OUTFILE=${SOURCEFILE%.scm}
        if ( bigloo -s -o $OUTFILE $SOURCEFILE ); then
            /bin/rm $OUTFILE.o
            shift
            ./$OUTFILE $@
        fi
        ;;
    *)
        echo Error: need a \*.scm file!
        ;;
esac
And the following Bigloo script: 

#! /usr/bin/env ./bgl
(module foo
   (main main))

(define (main argv)
   (print "foo: " argv))
When executed in the following way: 

$ chmod u+x foo.scm
$ ./foo.scm
  ⇥ foo: (./foo.scm)
The Bigloo module foo.scm will first be compiled and then executed. Of course, one may consider more complex compiler drivers where it is first checked that the module is not already compiled.