Not entirely serious: recursive lambda?

[Michael Tobis]

I came across the "japh" concept today and decided to do one of my
own, obviously, interpreting the 'p' somewhat loosely,

http://en.wikipedia.org/wiki/JAPH

but I'm not entirely satisfied with it:

####
# japh, for certain values of 'p'

f=lambda(r,N):N and f((" acdefijlmnopqrs tuv"[N%19]+r,N/19))or(r,N)
print f( ("",reduce(lamb da
c,m:c*95+''.joi n(map(chr,range (32,127))).inde x(m),
"!b)'1Mgh0z +fYQ]g::i^<&y~g)cnE-d=K&{GMNQ1gx+oo Y<~L##N'X]P2<@XYXwX3z",
0)))[0]

####

it bothers me that there are two statements. (In case you are
wondering what they do, it's all essentially about changing from base
95 to base 19. It's based loosely on this fine, simple recipe by Drew
Perttula which I have found to be useful on several occasions:

http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/111286

)

Anyway, I'd much prefer an even uglier japh like this:

# does not work
print (lambda(r,N):N and $$$$((" acdefijlmnopqrs tuv"[N%19]+r,N/
19))or(r,N))(
("",reduce(lamb da c,m:c*95+''.joi n(map(chr,range (32,127))).inde x(m),
"!b)'1Mgh0z +fYQ]g::i^<&y~g)cnE-d=K&{GMNQ1gx+oo Y<~L##N'X]P2<@XYXwX3z",
0)))[0]

but what would $$$$ be for an unnamed function to call itself?

I realize that lambda is something of an orphan and was arguably a bad
idea for anything besides obfuscation, but obfuscation is exactly my
purpose here. Can a lambda call itself without giving itself a name?
Google was not my friend on this one, and I suspect there is no
answer. Relax, I am not going to submit a PEP about it.

mt

[Miles]

Re: Not entirely serious: recursive lambda?

On Sat, Jul 19, 2008 at 10:43 PM, Michael Tobis <mtobis@gmail.c omwrote:

Can a lambda call itself without giving itself a name?

Kind of. There's a couple ways I know of.

The functional way, which involves the lambda receiving itself as an argument:

(lambda f: f(10, f))(lambda n, f: n and (sys.stdout.wri te("%d\n" % n)
or f(n-1,f)))

The stack frame examination way:

import sys, inspect, new
(lambda:sys.std out.write('recu rse\n') or
new.function(in spect.currentfr ame().f_code, globals())())()

The functional way is probably harder to grok unless you've studied
lambda calculus or had experience with "real" functional languages (I
haven't). For fun, try throwing a Y combinator in there.

-Miles

[Paul McGuire]

Re: Not entirely serious: recursive lambda?

On Jul 19, 11:49 pm, Miles <semantic...@gm ail.comwrote:

On Sat, Jul 19, 2008 at 10:43 PM, Michael Tobis <mto...@gmail.c omwrote:

Can a lambda call itself without giving itself a name?

>
Kind of.  There's a couple ways I know of.
>
The functional way, which involves the lambda receiving itself as an argument:
>
(lambda f: f(10, f))(lambda n, f: n and (sys.stdout.wri te("%d\n" % n)
or f(n-1,f)))
>
The stack frame examination way:
>
import sys, inspect, new
(lambda:sys.std out.write('recu rse\n') or
new.function(in spect.currentfr ame().f_code, globals())())()
>
The functional way is probably harder to grok unless you've studied
lambda calculus or had experience with "real" functional languages (I
haven't).  For fun, try throwing a Y combinator in there.
>
-Miles

Here is Michael Tobis's original program, using the functional
approach:

print (lambda f:f(("",reduce( lambda
c,m:c*95+''.joi n(map(chr,range (32,127))).inde x(m),
"!b)'1Mgh0z +fYQ]g::i^<&y~g)cnE-d=K&{GMNQ1gx
+ooY<~L##N'X]P2<@XYXwX3z",
0),f)))(lambda (r,N,f):N and f((" acdefijlmnopqrs tuv"[N%19]+r,N/
19,f))or(r,N,f) )[0]

Très assombri! (according to Babelfish...).

-- Paul

[Kay Schluehr]

Re: Not entirely serious: recursive lambda?

On 20 Jul., 04:43, Michael Tobis <mto...@gmail.c omwrote:

Can a lambda call itself without giving itself a name?

Sure, use a fixed point combinator. I've just added this recipe:

http://aspn.activestate.com/ASPN/Cookbook/Python/Recipe/576366

Google was not my friend on this one, and I suspect there is no
answer.

Even the Great Google can't help if you don't use the right
keywords ;)

[bearophileHUGS@lycos.com]

Re: Not entirely serious: recursive lambda?

Kay Schluehr:

Sure, use a fixed point combinator. I've just added this recipe:
http://aspn.activestate.com/ASPN/Coo.../Recipe/576366

Does it work?

Bye,
bearophile

[Kay Schluehr]

Re: Not entirely serious: recursive lambda?

On 20 Jul., 13:08, bearophileH...@ lycos.com wrote:

Kay Schluehr:
>

Sure, use a fixed point combinator. I've just added this recipe:
http://aspn.activestate.com/ASPN/Coo.../Recipe/576366

>
Does it work?
>
Bye,
bearophile

There are lots of informal derivations of the Y combinator on the web.
I used one a while ago and translated the result into Python. So if
there isn't an implementation bug it shall work.

I added two examples for illustration purposes.

[Fredrik Lundh]

Re: Not entirely serious: recursive lambda?

Michael Tobis wrote:

I realize that lambda is something of an orphan and was arguably a bad
idea for anything besides obfuscation, but obfuscation is exactly my
purpose here. Can a lambda call itself without giving itself a name?
Google was not my friend on this one, and I suspect there is no
answer. Relax, I am not going to submit a PEP about it.

gerson kurz' writings on lambdaization might be helpful (or not):

http://www.p-nand-q.com/python/lambdaizing_quicksort.html

http://www.p-nand-q.com/python/stupid_lambda_tricks.html

</F>

[Michael Tobis]

Re: Not entirely serious: recursive lambda?

Thanks all! What a remarkable set of answers, intelligent, thought
provoking and informative without exception.

Of course, now I can't use Paul's version; it hardly counts as a japh
if someone else wrote it! It is probably the closest to my original
vision, alas. Miles' second suggestion was the one I was fumbling
toward; I will study it. No spoilers please.

best
mt

[Paul McGuire]

Re: Not entirely serious: recursive lambda?

On Jul 22, 11:52 am, Michael Tobis <mto...@gmail.c omwrote:

Thanks all! What a remarkable set of answers, intelligent, thought
provoking and informative without exception.
>
Of course, now I can't use Paul's version; it hardly counts as a japh
if someone else wrote it! It is probably the closest to my original
vision, alas. Miles' second suggestion was the one I was fumbling
toward; I will study it. No spoilers please.
>
best
mt

Michael -

Sorry to spoil your fun - the concept of a recursive lambda was such
an interesting diversion, I couldn't resist! I'll try to restrain
myself next time.

-- Paul