Enchilada: the new kid on the block.
r_v_dalen — 2006-09-01 07:45:54
I've recently started a blog explaining the details of a new concatenative language
Enchilada.
Here are some statements to give you a taste of Enchilada:
- There are only two (sequence) types: expressions and lists.
- Expressions can be reduced at any level.
- Numbers follow naturally from lists.
Enchilada's most powerful operators are de-solve (!) and mul (*). Joy's quotations are not
included but are more or less replaced by lists.
Here are some juicy examples:
[2 3, 4 5] [+, *] * => [2 3 +, 4 5 +, 2 3 *, 4 5 *] =>
[5, 9, 6, 45]
[1, 2, 3, 4] \ % [+] * => [1, 2] [3, 4] % [+] * =>
[1 2, 3 4] [+] * => [1 2 +, 3 4 +] => [3, 7]
3 == [', ', ']
3 [1, 2] * => [1, 2, 1, 2, 1, 2]
Enchilada is still in design mode. I'm still pondering on the total number of primitives
operators to allow.
Feedback and suggestions are much appreciated!
See:
http://my.opera.com/rapido/blog/
William Tanksley, Jr — 2006-09-04 04:04:47
r_v_dalen <
r_v_dalen@...> wrote:
> I've recently started a blog explaining the details of a new concatenative language
> Enchilada.
I'm struggling my way through it. Looks nice and original. Fun reading.
> Enchilada is still in design mode. I'm still pondering on the total number of primitives
> operators to allow.
There was a paper on the basic primitives of Joy... Let's see, who
wrote that... Ah.
http://tunes.org/~iepos/joy.html
Honestly, I don't know whether it'll help you -- Enchilada is _very_ different.
Good luck!
-Billy
Robbert van Dalen — 2006-09-04 21:08:40
> I'm struggling my way through it. Looks nice and
> original. Fun reading.
Thanks. Hopefully I didn't leave anything to trip over
;).
> > Enchilada is still in design mode. I'm still
> pondering on the total number of primitives
> > operators to allow.
>
> There was a paper on the basic primitives of Joy...
> Let's see, who
> wrote that... Ah. http://tunes.org/~iepos/joy.html
Yes, I've read that paper (I've been lurking the
concatenative group for almost two years, so that's
why I know). It is very well written.
And yes, you can have a minimal set of operators, but
we should also care for efficiency. For instance,
Church numerals are mentioned in the paper. They are
pretty cool, but do not scale (theoretically) that
well in terms of big O.
> Honestly, I don't know whether it'll help you --
> Enchilada is _very_ different.
Do you think so? I think with some effort, Enchilada
can be mapped to Joy and vice versa.
Of course, the real difference will be performance,
because Enchilada has sequences build in as a
primitive.
>
> Good luck!
Thanks!
>
> -Billy
- Robbert
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
William Tanksley, Jr — 2006-09-05 17:57:20
Robbert van Dalen <
r_v_dalen@...> wrote:
> > I'm struggling my way through it. Looks nice and
> > original. Fun reading.
> Thanks. Hopefully I didn't leave anything to trip over
> ;).
I can't tell -- I'm still stumbling over my own shoelaces.
> And yes, you can have a minimal set of operators, but
> we should also care for efficiency. For instance,
> Church numerals are mentioned in the paper. They are
> pretty cool, but do not scale (theoretically) that
> well in terms of big O.
Oh, I agree. As you very well know, I'm big on efficiency :-). Now
that you've mentioned it... I'd like to know what plans you have for
efficiency. It's nice to be able to specify 2^n multiplications in a
single phrase, but it would be nice to not always have to perform 2^n
multiplications, list constructions, etc for every single phrase :-).
I'm a big fan of J and K -- can Enchilada be compared to them? It
seems like it could, if only there were a way to specify that only
some of the many possible terms of a given cartesian expansion needed
to be evaluated.
> > Honestly, I don't know whether it'll help you --
> > Enchilada is _very_ different.
> Do you think so? I think with some effort, Enchilada
> can be mapped to Joy and vice versa.
It certainly can; but it would be a waste (I almost typoed that as
"paste", more later) to fail to extract the interesting and unique
lessons Enchilada has to teach us, just because it's Turing complete.
> Of course, the real difference will be performance,
> because Enchilada has sequences build in as a
> primitive.
OTOH, it doesn't have numbers built in as a primitive... Is that a shortcoming?
> - Robbert
-Billy
Robbert van Dalen — 2006-09-05 19:19:48
> I can't tell -- I'm still stumbling over my own
> shoelaces.
Yeah I know what you mean. I'm still waiting for the
school bell to ring so I can finally play outside -not
to be bothered with mundane work.
> Oh, I agree. As you very well know, I'm big on
> efficiency :-). Now
> that you've mentioned it... I'd like to know what
> plans you have for
> efficiency. It's nice to be able to specify 2^n
> multiplications in a
> single phrase, but it would be nice to not always
> have to perform 2^n
> multiplications, list constructions, etc for every
> single phrase :-).
Right, I'd like to cut 2^n multiplications short, just
as anyone else. But my gut feeling says that it's not
going to happen soon.
> I'm a big fan of J and K -- can Enchilada be
> compared to them? It
> seems like it could, if only there were a way to
> specify that only
> some of the many possible terms of a given cartesian
> expansion needed
> to be evaluated.
That would be an interesting optimization. May be
http://www.nsl.com/papers/valence.htm can help us to
calculate the valence and charge of every possible
cartesian expansion.
> It certainly can; but it would be a waste (I almost
> typoed that as
> "paste", more later) to fail to extract the
> interesting and unique
> lessons Enchilada has to teach us, just because it's
> Turing complete.
Point taken. But I know where I come from: Joy has
teached me the most important lessons (do I hear the
bell ringing?)
> OTOH, it doesn't have numbers built in as a
> primitive... Is that a shortcoming?
Nah, not really. We can always (internally) flag lists
(with only empty sequences) as BigNumbers.
> -Billy
-Robbert
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com