corelinux-develop — Developer centric mail list for CoreLinux++

[Corelinux-develop] Library discussion

[Frank V. C. <fr...@co...>]

Christophe brought up a question about the Library Loader framework
implementation.

It may be useful to read this along side the design diagram

For the purposes of our discussion:

Library - This is really a registration point for users (developers) to
register (at run time) a loader. To register a loader you need two (2)
things: A Loader instance, and a LibraryType instance. Assuming there is
no conflict, the Library now has a Loader service which can begin work.
When a user calls the Library to create a LibraryInstance, they must
include two things, the name of the library, and the type of loader that
should handle the actual loading. At this point, a LibraryInstance is
given back to the caller.

Loader - Besides the physical resolution and loading of a library, the
Loader is a factory for LibraryInstances.

LibraryInstance - is the in memory object that represents, for all
intents and purposes the library that was loaded. The responsibility for
the LibraryInstance is to create, upon request, a LibraryObject. The
LibraryInstance interacts with a ObjectDefinitionRegistry to determine,
at run time, which type of LibraryObject derivation to instantiate. This
is done by the caller registering a series of LibraryObjectDefinitions
that describe what LibraryObject derivation should be instantiated upon
request. This works by using the key from the LibraryInstance as
associating LibraryObjectDefinitions to it.

ObjectDefinitionRegistry - an manager of the associations between a
library and the LibraryObjectDefinitions that it contains. 

LibraryObjectDefinition - a prototype which describes the interface and
behavior of the actual LibraryObject instance. These are a sort of Class
class, or meta-model which can be reasoned with at run time.

LibraryObject - is what the client actually interacts with to do
whatever work with the object that the library type has described. 

---- end

Because it is a framework, it is required that the developer extend a
number of the components to work together. We will actually be
implementing the first extension which will handle loading dynamic
shared libraries whose functions have a C calling convention. To do
this, we will provide:

FunctionLibraryType (LibraryType) - fairly vacuous type derivation.
FunctionLibraryLoader (Loader) - utilizes dlopen and dlclose to open
libraries.
FunctionLibrary (LibraryInstance) - exposes interface for getting and
returning ExecutionObjects.
ExecutionObjectDefinitions (LibraryObjectDefinition) - used to describe
ExecutionObjects
ExecutionObject (LibraryObject) - exposes the ability to "execute" with
arguments and return values.

What the user will be responsible for to utilize the FunctionLibrary
series is to create whatever ExecutionObjectDefinitions are needed at
run time to serve up useable objects.

Now, this is a big "small" step for CoreLinux++ in that we are getting
more into a framework type component, than the general patterns or IPC
objects to date. There is a few issues that need to be cleared up as
well, for example, like "HOW" do we describe arguments and return
semantics? I have done this before but in a big way, maybe there is a
better way.

-- 
Frank V. Castellucci
http://corelinux.sourceforge.net
OOA/OOD/C++ Standards and Guidelines for Linux
http://PythPat.sourceforge.net
Pythons Pattern Package

Re: [Corelinux-develop] examples

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> I forgot to mention that it would be good
> to keep the way the examples are named I mean the one containing the main()
> function because that is how the Makefile generation script(debian/genmake/pl)
>  knows how to build the executables.
> And don't duplicate filenames :) because everything is copied in the same
> directory
> 
> C.

Ok, no PDF and PS again. I have log file (although nothing seems clear).

Can you IRC 
irc.linux.com 
#sourceforge

?
That way I can describe what I see?

-- 
Frank V. Castellucci
http://corelinux.sourceforge.net
OOA/OOD/C++ Standards and Guidelines for Linux
http://PythPat.sourceforge.net
Pythons Pattern Package

[Corelinux-develop] examples

[Christophe Prud'h. <pru...@MI...>]

I forgot to mention that it would be good
to keep the way the examples are named I mean the one containing the main()
function because that is how the Makefile generation script(debian/genmake/pl)
 knows how to build the executables.
And don't duplicate filenames :) because everything is copied in the same
directory

C.

Re: [Corelinux-develop] cleanup

[Christophe Prud'h. <pru...@MI...>]

ok I update the spec and Makefile.am, now everything should be ok!
except the doc maybe

C.

Re: [Corelinux-develop] cleanup

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> >>>>> "Frank" == Frank V Castellucci <fr...@co...> writes:
>     Frank> Hmm. So your methodology is to do build then distclean? That's fine. And yes, it is good
> the make distclean is done in /usr/src/redhat/libcorelinux-0.4.26 not in your cvs copy
> the thing is that the make rpm create a dist and copy it in /usr/src/redhat/SOURCES
> (rpm requirement) the sources are untargzipped in /usr/src/redhat/BUILD

Everything but the resulting RPM files should be removed. Because it
does not do delta builds, as you pointed out, it is just a waste of
space to keep it around.

> 
>     Frank> that ANYTHING in /var/tmp is removed. I will keep looking after this build to see what
>     Frank> else so hold off for about 20 mins.  Is that ok?
> yes I will update in a minute
> 
> as for the doc for me it works fine. strange.

I'm running again with a log to see what the problem may be.
 

-- 
Frank V. Castellucci
http://corelinux.sourceforge.net
OOA/OOD/C++ Standards and Guidelines for Linux
http://PythPat.sourceforge.net
Pythons Pattern Package

Re: [Corelinux-develop] cleanup

[Christophe Prud'h. <pru...@MI...>]

>>>>> "Frank" == Frank V Castellucci <fr...@co...> writes:
    Frank> Hmm. So your methodology is to do build then distclean? That's fine. And yes, it is good
the make distclean is done in /usr/src/redhat/libcorelinux-0.4.26 not in your cvs copy
the thing is that the make rpm create a dist and copy it in /usr/src/redhat/SOURCES 
(rpm requirement) the sources are untargzipped in /usr/src/redhat/BUILD 

    Frank> that ANYTHING in /var/tmp is removed. I will keep looking after this build to see what
    Frank> else so hold off for about 20 mins.  Is that ok?
yes I will update in a minute

as for the doc for me it works fine. strange.

-- 
  Christophe Prud'homme                 |  Its name is Public Opinion.
  MIT, 77, Mass Ave, Rm 3-243           |  It is held in reverence.
  Cambridge MA 02139                    |  It settles everything.
  Tel (Office) : (00 1) (617) 253 0229  |  Some think it is the voice of God.
  Fax (Office) : (00 1) (617) 258 8559  |  -- Mark Twain
  http://augustine.mit.edu/~prudhomm    |
                          Following the hacker spirit

Re: [Corelinux-develop] cleanup

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
>     >> 2- cleanup
>     >>
>     >> I just added the rule it was commented sorry
> 
>     Frank> Checked in?
> 
> there is stuff in /var/tmp/libcorexxxxx and /usr/src/redhat/libcorelxxxxxx
> do you want both removed ?
> neither were removed
> now /var/tmp/libcorexxxx is removed
> and /usr/src/redhat/libcorexxx is cleanup (using make distclean)

Hmm. So your methodology is to do build then distclean? That's fine. And
yes, it is good that ANYTHING in /var/tmp is removed. I will keep
looking after this build to see what else so hold off for about 20 mins.
Is that ok?


-- 
Frank V. Castellucci

[Corelinux-develop] cleanup

[Christophe Prud'h. <pru...@MI...>]

    >> 2- cleanup
    >>
    >> I just added the rule it was commented sorry

    Frank> Checked in?

there is stuff in /var/tmp/libcorexxxxx and /usr/src/redhat/libcorelxxxxxx
do you want both removed ?
neither were removed 
now /var/tmp/libcorexxxx is removed 
and /usr/src/redhat/libcorexxx is cleanup (using make distclean)

comments ?
C.

Re: [Corelinux-develop] rpm bugs

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> >>>>> "Frank" == Frank V Castellucci <fr...@co...> writes:
> 
>     Frank> Christophe Prud'homme wrote:
>     >>
>     >> 1- the documentation missing files
>     >>
>     >> did you check in /usr/doc/libcorelinux-doc-0.4.26/ps /usr/doc/libcorelinux-doc-0.4.26/pdf
> 
>     Frank> They were empty. The actual build files were in the /latex directory but never made it to
>     Frank> gzip or rpm. I will run again and pipe output to file.
> I will investigate also
ok

> 
>     >> 2- cleanup
>     >>
>     >> I just added the rule it was commented sorry
> 
>     Frank> Checked in?
> Not yet
cool
> 
>     >> 3- shared library links
>     >>
>     >> corrected I included only libcl++.so.* and not libcl++.so* that's why it wasn't there
> 
>     Frank> Why? Don't -lcl look for libcl++.so ?
> at some point there is a rule saying what of kind files you want in certain location
> and I put this . between so and * therefore libcl++.so was not included
> now it is fixed
your a GOD!
> 
>     Frank> I'm saving up for a MP zeon machine.
> what is that ?
> zeon == xeon ?
heh, heh, yes Xeon

> if so we have a xeon here and I found that the xeon are much ado about nothing
> It is more expensive and the gain is not so clear

No, I have done benchmarks with UP and SMP with PIII and Xeon, with the
kind of workloads I have (lots of ephemeral (in memory) stuff) it has
about 25-40% improvement in throughput.

Frank

Re: [Corelinux-develop] rpm bugs

[Christophe Prud'h. <pru...@MI...>]

>>>>> "Frank" == Frank V Castellucci <fr...@co...> writes:

    Frank> Christophe Prud'homme wrote:
    >>
    >> 1- the documentation missing files
    >>
    >> did you check in /usr/doc/libcorelinux-doc-0.4.26/ps /usr/doc/libcorelinux-doc-0.4.26/pdf

    Frank> They were empty. The actual build files were in the /latex directory but never made it to
    Frank> gzip or rpm. I will run again and pipe output to file.
I will investigate also

    >> 2- cleanup
    >>
    >> I just added the rule it was commented sorry

    Frank> Checked in?
Not yet

    >> 3- shared library links
    >>
    >> corrected I included only libcl++.so.* and not libcl++.so* that's why it wasn't there

    Frank> Why? Don't -lcl look for libcl++.so ?
at some point there is a rule saying what of kind files you want in certain location
and I put this . between so and * therefore libcl++.so was not included 
now it is fixed

    Frank> I'm saving up for a MP zeon machine.
what is that ?
zeon == xeon ?

if so we have a xeon here and I found that the xeon are much ado about nothing
It is more expensive and the gain is not so clear

Re: [Corelinux-develop] rpm bugs

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> 1- the documentation missing files
> 
> did you check in
>  /usr/doc/libcorelinux-doc-0.4.26/ps
>  /usr/doc/libcorelinux-doc-0.4.26/pdf

They were empty. The actual build files were in the /latex directory but
never made it to gzip or rpm. I will run again and pipe output to file.

> 2- cleanup
> 
> I just added the rule
> it was commented sorry

Checked in?

> 3- shared library links
> 
> corrected I included only libcl++.so.* and not libcl++.so*
> that's why it wasn't there

Why? Don't -lcl look for libcl++.so ?

> I will update the cvs
> I just check the rpm building on my side
> 
> PS:
>  * the -j 2 option for make will be removed also(not everybody have a bi-pro)

I have one I just sold at a tag sale (2x133 PII).

I'm saving up for a MP zeon machine. 
 
> regards
> C.
> 
> --
>   Christophe Prud'homme                  |
>   MIT, 77, Mass Ave, Rm 3-243            |   If it doesn't work, force it.
>   Cambridge MA 02139                     |   If it breaks, it needed
>   Tel (Office) : (00 1) (617) 253 0229   |   replacing anyway.
>   Fax (Office) : (00 1) (617) 258 8559   |
>   http://augustine.mit.edu/~prudhomm     |
>                           Following the hacker spirit
> 
> _______________________________________________
> Corelinux-develop mailing list
> Cor...@li...
> http://lists.sourceforge.net/mailman/listinfo/corelinux-develop

-- 
Frank V. Castellucci
http://corelinux.sourceforge.net
OOA/OOD/C++ Standards and Guidelines for Linux
http://PythPat.sourceforge.net
Pythons Pattern Package

[Corelinux-develop] rpm bugs

[Christophe Prud'h. <pru...@MI...>]

1- the documentation missing files 

did you check in 
 /usr/doc/libcorelinux-doc-0.4.26/ps
 /usr/doc/libcorelinux-doc-0.4.26/pdf 

I have them 

2- cleanup

I just added the rule
it was commented sorry

3- shared library links

corrected I included only libcl++.so.* and not libcl++.so*
that's why it wasn't there

I will update the cvs 
I just check the rpm building on my side

PS:
 * the -j 2 option for make will be removed also(not everybody have a bi-pro)

regards
C.

-- 
  Christophe Prud'homme                  |
  MIT, 77, Mass Ave, Rm 3-243            |   If it doesn't work, force it.
  Cambridge MA 02139                     |   If it breaks, it needed
  Tel (Office) : (00 1) (617) 253 0229   |   replacing anyway.
  Fax (Office) : (00 1) (617) 258 8559   |
  http://augustine.mit.edu/~prudhomm     |
                          Following the hacker spirit

[Corelinux-develop] doxygen

[Christophe Prud'h. <pru...@MI...>]

Hi all,

it seems that the patches I sent to doxygen have been applied 
so if you want to get the nice pdf manual
get the latest doxygen cvs stuff !

regards
C.

-- 
  Christophe Prud'homme                 |
  MIT, 77, Mass Ave, Rm 3-243           |  The Second Law of Thermodynamics:
  Cambridge MA 02139                    |  If you think things are in a mess
  Tel (Office) : (00 1) (617) 253 0229  |   now, just wait!
  Fax (Office) : (00 1) (617) 258 8559  |  -- Jim Warner
  http://augustine.mit.edu/~prudhomm    |
                          Following the hacker spirit

[Corelinux-develop] Packages category added to BugTracking

[Frank V. C. <fr...@co...>]

I added the category to further delineate the defects.

C., sorry for the messages being updated, I added the category after the
defects and am going back and re-categorizing.
-- 
Frank V. Castellucci
http://corelinux.sourceforge.net
OOA/OOD/C++ Standards and Guidelines for Linux
http://PythPat.sourceforge.net
Pythons Pattern Package

Re: [Corelinux-develop] midnight is good to announce stuff

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> does it appear when you go in doc/
> and type make ref ?
> 
> I cvs updated corelinux/doc there is a new corelinux.cfg.in
> which should correct the problem!
> it seems that we do not have the same latex installation
> I removed the package html as an extra package!
> 
> just type make in corelinux/doc to regenerate the corelinux.cfg
> and go in latex to type make pdf to see if the problem is still there
> if it is ok then the package creation should be fine
> 
> C.

Ok, that fixed the rpm builds! I will now be testing install/uninstall
and let you know if anything crops up!

-- 
Frank V. Castellucci

Re: [Corelinux-develop] midnight is good to announce stuff

[Christophe Prud'h. <pru...@MI...>]

does it appear when you go in doc/
and type make ref ?


I cvs updated corelinux/doc there is a new corelinux.cfg.in
which should correct the problem!
it seems that we do not have the same latex installation
I removed the package html as an extra package!

just type make in corelinux/doc to regenerate the corelinux.cfg
and go in latex to type make pdf to see if the problem is still there
if it is ok then the package creation should be fine 

C.

-- 
   Christophe Prud'homme                  |
   MIT, 77, Mass Ave, Rm 3-243            |   I respect faith, but doubt is
   Cambridge MA 02139                     |   what gives you an education.
   Tel (Office) : (00 1) (617) 253 0229   |   -- Wilson Mizner
   Fax (Office) : (00 1) (617) 258 8559   |
   http://augustine.mit.edu/~prudhomm     |
                          Following the hacker spirit

Re: [Corelinux-develop] midnight is good to announce stuff

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> The rpm packaging is done :)
> 
> the examples appeared to be easy using the scripts I used for debian
> I will upload the deb/rpm packages to corelinux.sourceforge.net/packages
> so that you can give some feedback
> 
> a corelinux.spec.in file will appear in your cvs
> to generate the packages you have to do:
> 
> > autorun.sh
> > configure --> will generate corelinux.spec [1]
> > make rpm  --> you must have read-write access /usr/src/redhat or be root
> 
> then take a coffee or a tea and relax, it takes some time !
> 
> Footnotes:
> [1]  automatic versioning Frank! You will be happy, this one more
>      to not worry about when releasing corelinux :)

This captures the error at which point the 'make rpm' hangs waiting for,
yechh, human intervention.
-- 
Frank V. Castellucci

Re: [Corelinux-develop] midnight is good to announce stuff

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
>     Frank> Cool stuff, I will download and build as well as install and test. So when the smoke
>     Frank> clears, I will be genning the rpm packages, tarball, PDF, DVI, TeX, and you will be doing
>     Frank> the deb packages until we can get the compile farm machine to outfit the whole thing yes?
> yes!
> I forsee one problem:
> the reference generated by doxygen
> I build the manual during the package creation so if you do not have
> doxygen/latex installed that we'll be a problem :(
> 
> regards
> C.

I was having some problem with 'make rpm'. I am running again (I
installed doxygen) and will report back if a problem. 

-- 
Frank V. Castellucci

Re: [Corelinux-develop] midnight is good to announce stuff

[Christophe Prud'h. <pru...@MI...>]

    Frank> Cool stuff, I will download and build as well as install and test. So when the smoke
    Frank> clears, I will be genning the rpm packages, tarball, PDF, DVI, TeX, and you will be doing
    Frank> the deb packages until we can get the compile farm machine to outfit the whole thing yes?
yes!
I forsee one problem:
the reference generated by doxygen
I build the manual during the package creation so if you do not have 
doxygen/latex installed that we'll be a problem :(

regards
C.

-- 
Christophe Prud'homme                |
MIT, 77, Mass Ave, Rm 3-243          | He that breaks a thing to find out
Cambridge MA 02139                   | what it is has left the path of wisdom.
Tel (Office) : (00 1) (617) 253 0229 | -- J.R.R. Tolkien
Fax (Office) : (00 1) (617) 258 8559 |
http://augustine.mit.edu/~prudhomm   |
                          Following the hacker spirit

Re: [Corelinux-develop] midnight is good to announce stuff

[Frank V. C. <fr...@co...>]

Christophe Prud'homme wrote:
> 
> The rpm packaging is done :)
> 
> the examples appeared to be easy using the scripts I used for debian
> I will upload the deb/rpm packages to corelinux.sourceforge.net/packages
> so that you can give some feedback
> 
> a corelinux.spec.in file will appear in your cvs
> to generate the packages you have to do:
> 
> > autorun.sh
> > configure --> will generate corelinux.spec [1]
> > make rpm  --> you must have read-write access /usr/src/redhat or be root
> 
> then take a coffee or a tea and relax, it takes some time !
> 
> Footnotes:
> [1]  automatic versioning Frank! You will be happy, this one more
>      to not worry about when releasing corelinux :)

:->

Cool stuff, I will download and build as well as install and test. 
So when the smoke clears, I will be genning the rpm packages, tarball,
PDF, DVI, TeX, and you will be doing the deb packages until we can get
the compile farm machine to outfit the whole thing yes?

-- 
Frank V. Castellucci

[Corelinux-develop] Feature 110459 [packaging] new redhat packaging system

[Christophe Prud'h. <pru...@MI...>]

The rpm files are in /home/groups/corelinux/packages/debian
The cvs is updated

enjoy ![1]

TODO:
-----
o document the ACCheckCoreLinux feature in Developper manual
o create a README.packages to document package creation
  and packages contents
o eventually bug fixes in packaging system

good night
C.


Footnotes: 
[1]  I'll try to contact mandrake to include libcorelinux in there next
     release using their cooker system 
     http://www.linux-mandrake.com/en/cookerdevel.php3
     I dunno if there is such a thing with redhat or suse

-- 
  Christophe Prud'homme                 |  A classic is something that
  MIT, 77, Mass Ave, Rm 3-243           |  everyone wants to have read and
  Cambridge MA 02139                    |  nobody wants to read.
  Tel (Office) : (00 1) (617) 253 0229  |  -- Mark Twain,
  Fax (Office) : (00 1) (617) 258 8559  |  "The Disappearance of Literature"
  http://augustine.mit.edu/~prudhomm    |
                          Following the hacker spirit

[Corelinux-develop] midnight is good to announce stuff

[Christophe Prud'h. <pru...@MI...>]

The rpm packaging is done :)

the examples appeared to be easy using the scripts I used for debian
I will upload the deb/rpm packages to corelinux.sourceforge.net/packages
so that you can give some feedback

a corelinux.spec.in file will appear in your cvs
to generate the packages you have to do:

> autorun.sh
> configure --> will generate corelinux.spec [1]
> make rpm  --> you must have read-write access /usr/src/redhat or be root

then take a coffee or a tea and relax, it takes some time !

Footnotes: 
[1]  automatic versioning Frank! You will be happy, this one more
     to not worry about when releasing corelinux :)

Re: [Corelinux-develop] rpm packages

[Hans - D. <dul...@eg...>]

On Fri, 28 Jul 2000, Christophe Prud'homme wrote:

> Date: Fri, 28 Jul 2000 23:38:31 -0400 (EDT)
> From: Christophe Prud'homme <pru...@MI...>
> Reply-To: cor...@li...
> To: cor...@li...
> Subject: [Corelinux-develop] rpm packages
> 
> We are getting closer to rpm packaging
> now we have:
> o libcorelinux
> o libcorelinux-dev
> o libcorelinux-doc
> 
> TODO:
> -----
> o libcorelinux-examples will be more painful
> 
> 
> Do you guys know how to not restart a rpm building from scratch?

Not me.

> each time it erases everything and restarts, so debugging/implementing
> the spec file is really painful 
> Fortunately I have a dual pentium III 750 so the -j 2 option of make is 
> really a winner, but that's brutal force
> 
> is there anything like time-stamp? or being able to restart the process
> at a certain point?
> help would be much appreciated :)
> I can't believe there is nothing like that
> 
> one more question(I am not very fluent in redhat systems)
> where each of the packages should go, I mean which categories?
> 
> regards
> C.
> 
> 

-- 
Hans Dulimarta, Ph.D.             dulimart@[egr.msu.edu, computer.org]
Visiting Research Associate           http://www.egr.msu.edu/~dulimart   
                           Ph: (517)432-7589  Fax2Email: (760)281-7691
Elec. & Comp. Engg., Michigan State University, East Lansing, MI 48824

[Corelinux-develop] rpm packages

[Christophe Prud'h. <pru...@MI...>]

We are getting closer to rpm packaging
now we have:
o libcorelinux
o libcorelinux-dev
o libcorelinux-doc

TODO:
-----
o libcorelinux-examples will be more painful


Do you guys know how to not restart a rpm building from scratch?
each time it erases everything and restarts, so debugging/implementing
the spec file is really painful 
Fortunately I have a dual pentium III 750 so the -j 2 option of make is 
really a winner, but that's brutal force

is there anything like time-stamp? or being able to restart the process
at a certain point?
help would be much appreciated :)
I can't believe there is nothing like that

one more question(I am not very fluent in redhat systems)
where each of the packages should go, I mean which categories?

regards
C.

-- 
  Christophe Prud'homme                  |
  MIT, 77, Mass Ave, Rm 3-243            |   If it doesn't work, force it.
  Cambridge MA 02139                     |   If it breaks, it needed
  Tel (Office) : (00 1) (617) 253 0229   |   replacing anyway.
  Fax (Office) : (00 1) (617) 258 8559   |
  http://augustine.mit.edu/~prudhomm     |
                          Following the hacker spirit

Re: [Corelinux-develop] status report

[Christophe Prud'h. <pru...@MI...>]

>>>>> "Frank" == Frank V Castellucci <fr...@co...> writes:

    Frank> Whenever the state of a defect/feature changes, you get e-mail notification if you are
    Frank> the handler or person who submitted.
yes I knew that
    Frank> But, do not CLOSE, use Fixed instead so I have something available to remind me of what
    Frank> goes into the ChangeLog when we do across the board releases.
Ok

I think there is a bug in sourceforge when submitting a bug
if you try to assign it to someone it won't work 
you have to submit, edit it and reassign it

C.

-- 
  Christophe Prud'homme                 |  A classic is something that
  MIT, 77, Mass Ave, Rm 3-243           |  everyone wants to have read and
  Cambridge MA 02139                    |  nobody wants to read.
  Tel (Office) : (00 1) (617) 253 0229  |  -- Mark Twain,
  Fax (Office) : (00 1) (617) 258 8559  |  "The Disappearance of Literature"
  http://augustine.mit.edu/~prudhomm    |
                          Following the hacker spirit