Re: [Slib-discuss] Incompatibilities Between Structures of SLIB and Gambit

[Aubrey Jaffer]

 | Date: Mon, 25 Jun 2012 17:48:44 -0500
 | From: HN <address@hidden>
 | 
 | The "structure" type (section 3.7 of SLIB documentation) and its 
 | Gambit's counterpart have the following differences (in order of 
 | decreasing importance):
 | 
 |    1. SLIB uses an extra pair of parentheses around "name field" while
 |       Gambit 4.0 does not, e.g.
 |           * SLIB: (define-structure (name field ...))
 |           * Gambit: (define-structure name field ...)
 | 
 |    2. The modifier's name is documented in SLIB as 'name-field-set!',
 |       but the example that follows and the implemented name is
 |       'set-name-field!' ("set" is put first instead of last).  Gambit
 |       documents and implements the former.
 | 
 |    3. SLIB's version allows for default values while Gambit does not.
 |       However, this is not mentioned in SLIB's document.
 | 
 | The first issue causes portability between SLIB and Gambit's code.  We
 | can change the source code of SLIB's "structure.scm" to match
 | Gambit's, or we can comment out this feature in "gambit.init" to
 | follow the SLIB's way.

The SLIB version was written in 1992 by R. Kent Dybvig.  Can someone
familiar with syntax-case macros modify slib/structure.scm to accept
unparenthesized arguments?

Even better would be to modify it to accept both a parenthesized
argument and unparenthesized arguments, to be compatible with existing
code.

 | The second can only be fixed by changing SLIB's file "structure.scm".

I have fixed this; the development version is updated:
http://groups.csail.mit.edu/mac/ftpdir/users/jaffer/slib.zip

Also, the CVS repository is updated:
https://savannah.gnu.org/cvs/?group=slib

It might be better to have it generate both old and new function
names for backward compatibility.

 | The third involves documenting this feature.

Can you supply text for the documentation?