gcl-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Gcl-devel] steps toward ANSI CL compatibility


From: Vadim V. Zhytnikov
Subject: [Gcl-devel] steps toward ANSI CL compatibility
Date: Sun, 02 Jun 2002 23:22:44 +0300

Greetings!

I've been playing with gcl the ansi-test suite from clocc.
Quickly I realized that at present gcl has several very important

ANSI CL incompatibilities.  Some of them must be fixed even
before running ansi-test with some sensible and useful result.

First of all - basic lisp package names.  gcl follows
old practice and defines two basic packages LISP and USER.
According to ANSI CL these packages must be named
COMMON-LISP with alias CL and COMMON-LISP-USER
with alias CL-USER.  The most important is the COMMON-LISP
package since standard imposes very severe restrictions on it.
This package must export all defined by the CL standard symbols
and must not have any other implementation dependent external
symbols.  LISP package is permitted as well for the sake of
compatibility with older implementations.  LISP package may
have implementation dependent extra external symbols.
So it seems to me that we should create new package
COMMON-LISP=CL
and import  in it all required symbols from LISP (this is how
CLtL2
recommends to resolve the problem).
COMMON-LISP-USER package is not so restrictive - it should
use COMMON-LISP but may have other symbols. So maybe it
is enough to make single package with three aliases
COMMON-LISP-USER=CL-USER=USER.

PCL (portable CLOCC implementation) and clcs (CONDITIONS package)

must be definitely incorporated in GCL. These packages are part
of the ANSI CL
standard and CL package must use them.  I've successfully
compiled pcl
and clcs and build new gcl image with pcl and clcs loaded.
The only peculiarity is that several symbols from clcs must be
shadowing
imported (they are listed in package.lisp).  I did it manually
but this
procedure must be incorporated in the standard configure/make
routine.

Best wishes,

Vadim

--
     Vadim V. Zhytnikov

      <address@hidden>
    <address@hidden>
     <address@hidden>






reply via email to

[Prev in Thread] Current Thread [Next in Thread]