ooc-compiler

Re: [ooc-compiler] Lightning Oberon anyone?

[dirk.muysers <dir...@li...>]

Michael van Acken <mi...@de...> wrote:

| It looks like there are still people at the ETH Zürich tinkering with
| programming language design: http://www.oberon.ethz.ch/lightning/
|
| Is anyone interested in doing an implementation of LO in the OOC
| framework?
|
| -- mva

How would he possibly do that?

1. LO is intimately bound to a new universal dynamic core language support
library that has not yet been released by Microsoft. He would need to
discuss a special non-disclosure agreement with MS in order to have access
to that library. Only a couple of universities and corporations are a part
of the ballgame so far.

2. LO is an ad-hoc design for that experience, viz. to prove a point (Oberon
being capable to play in the upper league), and its design is far from
having reached a state of maturity. While its approach to component design
is a step in the right direction (and one of the reasonable ways to
implement the Lagoona ideas --the Lagoona WEB pages BTW are no longer
accessible--, but then, has it really all the features required for
programming at large?

3. The biggest drawback of LO is its tight integration with Microsoft's .NET
platform. The design could indeed be elaborated in a more platform-agnostic
direction by having its own run-time library, which is quite a mouthful. In
any case, as I recently said in c.l.o., the Oberon-2 report is now old
enough to deserve a revision, in order to integrate the goodies of its
offsprings such as Component Pascal, Active Oberon etc., otherwise we will
land in a total mess.

ETH, being a university and research institute, has no particular interest
in Oberon becoming an industrial tool.

OOC is an open-software compiler project, maybe we should need an
open-software Oberon language project, kind of GNU/Oberon, in order to make
things advance.

OOC also badly needs native or at least assembler (NASM?) code generators in
order to become more usable.

[ooc-compiler] Lightning Oberon anyone?

[Michael v. A. <mi...@de...>]

It looks like there are still people at the ETH Zürich tinkering with
programming language design: http://www.oberon.ethz.ch/lightning/

Is anyone interested in doing an implementation of LO in the OOC
framework?

-- mva

[ooc-compiler] First Release libxml v2

[Michael v. A. <mi...@de...>]

Tarballs of the XML parser sources and test cases can be downloaded
from http://sourceforge.net/project/?group_id=3539.  Under the module
"libxml" in the "File Releases" section there are two bzip2 tar files.
The smaller holds the source files, the other the test cases.

The parser should be functionally equivalent to the old one.  All
modules are under the LGPL license.  Check the included README for
more details.

-- mva

[ooc-compiler] New XML Parser on SourceForge CVS

[Michael v. A. <mi...@de...>]

[As far as I know, ooc-list has not been moved to the SourceForge
 mailing list yet.  Nevertheless, I am only posting this to the
 SourceForge mailing list.]

I've uploaded the sources of the new XML parser to the SourceForge CVS
repository.  The sources include the XML parser itself, encoder and
decoder for the most prominent Unicode encodings, an URI library, and
some auxiliary OS level functions.

At the moment, the parser succeeds for all well-formedness tests I can
lay may hands on (which are approx. 750 different test cases).  I'm
currently working on the interface to provide all the information
(core and periphereal) of the XML Information Set to the application.
In the future, I will add support for namespaces (near future) and
validation (far or very far future).  Compared to the old XML parser,
this implementation's design is simpler and the new parser is much
faster.  Besides, it is much more widely tested and will be compliant
to the XML 1.0 spec and the current errata list in all points.
Another point some people will like: the sources are published under
the LGPL, _not_ GPL.


This is development stuff and I expect it is only interesting to a
select few.  To get the goodies, check the instructions at

  https://sourceforge.net/cvs/?group_id=3539

under "Anonymous CVS Access".  The module name is "libxml".  If you
only want to browse the source code, 

  http://cvs.sourceforge.net/cgi-bin/cvsweb.cgi/libxml/?cvsroot=ooc

is probably the interface of your choice.


There is no configure magic, which means you must add the directory to
the compiler's search path yourself (libxml/src must be included for
*.Mod and *.c).  To run the test cases, grab

  http://pweb.de.uu.net/vanacken.do/ooc/oo2c-patches/oo2c_32-1.5.2-1.5.2.9.diff.gz

This compiler patch fixes a glitch in the error reporting of module
`Files' that breaks some of the tests.  Set the environment variable
OOC to point to your oo2c compiler.  If your diff is not GNU, remove
the option "-u" in libxml/tests/Makefile.  Then, do "make test" in the
root directory.  You should see several hundred "ok" messages, and
_no_ errors.  (I'm not sure what happens if your "make" is not of the
GNU persuasion.)

Have fun.

-- mva

[ooc-compiler] Test: SourceForge Mailing List (again)

[Michael v. A. <mi...@de...>]

This is our second attempt to get a mailing list on SourceForge up and
running.  The previous attempt with ooc-general failed because SF
seems to use the "-general" suffix for a general umbrella list.

I've created another list, ooc...@li..., and
subscribed all recipients of ooc...@in... to it.
To unsubscribe, or to change your subscription parameters, please use 
  http://mail1.sourceforge.net/mailman/listinfo/ooc-compiler
You will find your entry under "Visit Subscriber List".  For
modifications, you'll have the system send you your password first (a
side effect of the bulk subscription).


Anyway, this mail is sent to both mailing lists.  That is, you should
receive two copies of this mail.  Please complain if this doesn't work
as advertised.  Otherwise, ooc-list will be aliased to ooc-compiler in
the very near future.

-- mva