[M3devel] JIT [WAS: Google Benchmark - anyone interested in an Modula 3 version?]

Daniel Alejandro Benavides D. dabenavidesd at yahoo.es
Tue Jul 5 23:38:01 CEST 2011


Hi all:
I see now, the reason was explained by Luca Cardelli on his "recent" interview, they asked to come with something of practical use, yet they had pressure, to retrieve some good quality work of good practical results, that become the Theory of Objects (with Abadi, see [3]) on:
http://citeseer.ist.psu.edu/viewdoc/summary?doi=10.1.1.34.4024

But the theory on Baby Modula-3 was the idealization they had on Modula-3 typing system, unfortunately more work on that got that direction, towards the final years on the laboratory was somewhat cut, that explains why Larch project wasn't continued there after initial works (fortunately continued though Greg Nelson ESC work, which is by itself more practical).
And yes, I believe this is path for a good structured at least terms comprehension of the language (I mean whenever I take foreign language, i.e English class i must study phonetics, semantics and grammar, well at least that's is what I was told by a former German language undergraduate student, specially grammar, which is for us specially hard, maybe be not the case for say you know, almost anyone else), so in this terms this is good to know for our future culprits, similarly we could argue for i.e us Computer scientists, the most valuable experience using Modula-3 was in Cambridge (and I have seen real good Modula-3 seminal work coming from there as well as other ones as well)
Abadi and Cardelli and Matthes created a theory to create extensible syntax and use it  in a DB realization of Modula-3:

http://www.hpl.hp.com/techreports/Compaq-DEC/SRC-RR-121.pdf

I guess this could be the way to go, besides the seminal work in denotational (to express Modula-3 correctness in baby Modula-3) is available for free from its author:
http://people.cis.ksu.edu/~schmidt/text/densem.html

I believe we could make a point on this results preliminary but very important for us to depart from the rest of approaches, or at least have our own version (Baby Modula-3 is not BTW implemented as for what I know about, I wish could be proved to be correct, but I don't know neither, could be interpreted as it is, so I haven't seen it).

That said, there is even more, some people went through the work or recapitulating the Theory of Objects in terms of denotational semantics

http://www.cl.cam.ac.uk/~jg531/mypubl/pre-PhD/direc-WOOD.pdf

And a way to unify them based on recursive types, or say a way to treat programming language type back and forth (note I didn't say types if so, this is not the case), one of its authors wrote the most used book on "Types for Programming languages":, the former [2]:

http://citeseer.ist.psu.edu/viewdoc/summary?doi=10.1.1.17.5855

Perhaps this is the way an actual Baby Modula-3 implementation might become a good vehicle to explain programming language semantics (maybe not programming languages at all, just their semantics).
If somebody agrees on that as Cardelli said to be this the next wave in his interview on Modula-3, let's bring it on www-baby.modula3.com for instance

Let mew know your thoughts, thanks in advance

[1] P. Robinson, “From ML to C via Modula-3 an approach to teaching programming,” 1994.

[2] M. Hofmann and B. Pierce, “A Unifying Type-Theoretic Framework for Objects,” 1993.

[3] M. Abadi and L. Cardelli, “A Semantics of Object Types,” PROC. IEEE SYMPOSIUM ON LOGIC IN COMPUTER SCIENCE, p. 332--341, 1994.


--- El mar, 5/7/11, Daniel Alejandro Benavides D. <dabenavidesd at yahoo.es> escribió:

> De: Daniel Alejandro Benavides D. <dabenavidesd at yahoo.es>
> Asunto: Re: [M3devel] JIT [WAS: Google Benchmark - anyone interested in an Modula 3 version?]
> Para: m3devel at elegosoft.com, "Hendrik Boom" <hendrik at topoi.pooq.com>
> Fecha: martes, 5 de julio, 2011 13:57
> Hi all:
> yes, in that domain yes, but in the many ways of solving a
> problem, computers have a rather limited extent, therefore
> you can't expect such facilities, but for things like
> automating algorithm comprehension, this is a task of
> algorithmic learning.
> A way of this ahs been proposed for a constraint language
> (e.g Juno),:
> http://books.google.com/books?id=m-BQAAAAMAAJ&q=juno+algorithmic+learning&dq=juno+algorithmic+learning&hl=en&ei=R1oTTqStJ-Tg0QHf0sTHDg&sa=X&oi=book_result&ct=result&resnum=1&ved=0CCkQ6AEwAA
>  
> Well that said, the idea behind of this is how to optimize
> the program, but to define a good type system implementable
> is important too, is good to have many kernel for doing
> that. Abadi-Cardelli is a good one, but certainly not the
> only one, Abadi Baby Modula-3 is yet another one, I can't
> understand, it should be be in the Modula-3 research
> reports, or in the bibliography, I will see, how about
> that.
> 
> Thanks in advance
> 
> --- El mar, 5/7/11, Hendrik Boom <hendrik at topoi.pooq.com>
> escribió:
> 
> > De: Hendrik Boom <hendrik at topoi.pooq.com>
> > Asunto: Re: [M3devel] JIT [WAS: Google Benchmark -
> anyone interested in an Modula 3 version?]
> > Para: m3devel at elegosoft.com
> > Fecha: martes, 5 de julio, 2011 12:34
> > On Tue, Jul 05, 2011 at 12:08:07PM
> > -0500, felipe valdez wrote:
> > > On Tue, Jul 5, 2011 at 11:50 AM, Hendrik Boom
> <hendrik at topoi.pooq.com>wrote:
> > > 
> > > > Ideally, the computer should do about the
> same
> > amount of type
> > > > inferencing that a human reader would easily
> do
> > when reading
> > > > the program.  Note that I said
> "reader", not
> > "writer".
> > > >
> > > >
> > > but how far are we from this Ideal, really?
> > > can these rules be automated?
> > 
> > The hard part here is figuring out what human readers
> can
> > do.  It's 
> > important to match the human reader and not to go
> beyond
> > his abilities.  
> > Because then the writer (who will probably understand
> his
> > program bettr 
> > than a reader) will get feedback from his type-checker
> when
> > the code 
> > gets too obscure for a reader.
> > 
> > -- hendrik
> >
> 



More information about the M3devel mailing list