We are getting ready to release our ILU system, and this message is
to solicit alpha testers for our first outside-of-PARC release.
What's ILU?
It stands for Inter-Language Unification, and it is an attempt at
providing interface technology for application programs in a language
neutral fashion. It does this by requiring interfaces to be written
in a neutral language, called ISL, for Interface Specification
Language, and, for each supported programming language, includes tools
for generating language-specific versions of the specified interface,
and libraries for supporting these language-specific interfaces.
Those familiar with the canonical approach to supporting RPC on UNIX
platforms will recognize this approach as being quite similar to that
used by various RPC toolkits. Exactly so, and ILU will indeed serve
to combine program fragments either in a single address space, or, via
RPC, over multiple address spaces. However, it does not dictate the
use of any specific RPC protocol for these distributed applications,
but is engineered to work with various RPC protocols. (We currently
have mappings to Sun RPC and Xerox Courier.) Similarly, it does not
dictate any specific underlying transport such as TCP/IP, but can work
with various reliable byte-stream transport mechanisms. (We currently
have mappings to TCP/IP and the XNS Sequenced Packet Protocol.) The
mappings to standard RPC protocols are such that some subset of
existing services can be used directly from ILU without modification
-- those for which an ISL interface description can be written.
The concepts supported in the ISL interface specification language
include
- namespaces (non-nested)
- 16, 32, and 64 bit integers, signed and unsigned (and bytes)
- 32, 64, and 128 bit floating point numbers
- ASCII and Unicode characters (and strings)
- boolean and pointer values
- structs (records), tagged unions, multi-dimensional arrays
- sequences (lists)
- object types (multiple-inheritance)
- object methods (synch/asynch) (result caching/non-caching)
- exceptions
- garbage collection
We currently have (working) mappings of these concepts to 5 languages:
Modula-3 (DEC SRC), Common Lisp (Franz Allegro 4.1), C++ (several
compilers, thought to be generic), FORTRAN (Sun f77), and ANSI C
(again thought to be generic). Mappings use an object-oriented
approach throughout; we add a simple object system to C and FORTRAN,
and extend the object system in Modula-3 to support multiple
inheritance. The C mapping conforms as much as possible to the C
mapping specified by the Object Management Group's Common Object
Request Broker Architecture, for their IDL. Our ISL parser will also
parse a subset of OMG IDL (disallowing any IDL "#include"s which
cannot be turned into ISL "IMPORT" statements; disallowing the more
baroque IDL union specifications; disallowing any IDL statements
outside the scope of a "module" statement; disallowing any definition
of anything but member functions or attributes inside an IDL
"interface" statement; disallowing the IDL "object" type; disallowing
the IDL "typecode" type).
The system is written in ANSI C on Sun SunOS 4.1.3 workstations. We
believe that it will port fairly easily to other operating systems,
and to other versions of UNIX. We have ported our kernel library and
our C++ support to the Macintosh, as a test. The system will be
released publicly later this year, under a copyright which should
freely allow use in almost any situation.
So what's an Alpha Tester?
We'd like to find *a small number* of alpha testers interested in 1)
porting the Lisp support to Lucid CL, CMU CL, and Mac CL; 2)
exercising the support for the various languages, and submitting bug
reports; 3) porting the kernel and tools to MS-Windows; 4) porting the
kernel and tools and support for various languages to different UNIX
platforms; 5) providing RPC support for OSF DCE. We intend to gather
together these various pieces of work and re-release this with our
beta test version. If you'd like to participate, please send me a
message with some description of why you in particular would be
wonderfully qualified to test/port this.
Bill
--
Bill Janssen ·······@parc.xerox.com (415) 812-4763
Xerox Palo Alto Research Center FAX: (415) 812-4777
3333 Coyote Hill Road, Palo Alto, California 94304
In article <·····················@holmes.PARC.Xerox.Com> ·······@PARC.Xerox.Com (Bill Janssen) writes:
! What's ILU?
! It stands for Inter-Language Unification, and it is an attempt at
! providing interface technology for application programs in a language
! neutral fashion. It does this by requiring interfaces to be written
! in a neutral language, called ISL, for Interface Specification
! Language, and, for each supported programming language, includes tools
! for generating language-specific versions of the specified interface,
! and libraries for supporting these language-specific interfaces.
Isn't this the same thing that CORBA is trying to do? If so, why
should one use ILU and not CORBA?
--
.................. .................. ..................
. Ari Huttunen . . Sather . . Haskell .
. . . . _ . .
. . OS/2 . . Godel . .
....... .................. .................. ...........