[Haskell-cafe] Re: New slogan for haskell.org

[Yitzchak Gale]

Jerzy Karczmarczuk wrote:
> ...don't forget that itertools have been introduced in the
> version 2.3! ...But we were speaking about
> map and filter, and this is 1994... So, perhaps some concrete references
>  - if given - should be tested against the historical truth.

OK, OK. You want to make me work.

In May 1994, there was the following post by Ken Manheimer in which he
praises the functional features of Python:

http://www.python.org/search/hypermail/python-1994q2/0698.html

That invoked the famous response by Guido in which he
regrets having introduced some of those features:

http://www.python.org/search/hypermail/python-1994q2/0705.html

Here are some excerpts:

"Really, I don't understand all the fuzz about lambda, and I don't see
why its introduction made functions any more first class than they
already were in Python... Also notice that Python's map, filter and
reduce are quite powerless compared to the same functions in the
average functional programming system -- they don't work when applied
to infinite sequences since they use eager evaluation."

Can we guess which "average functional programming system"
he was thinking of? Hint: it uses lazy evaluation.

Guido is clearly not rejecting functional influences
on Python, he is supporting them. But he feels that
these specific instances do not fit in.

The thread then continues with a wide-ranging discussion of
functional versus imperative programming, including CPS style
(intertwined with the original topic, which was Python's layout
vs. C-style blocks, where Haskell is also mentioned).
This discussion makes it clear that all those who worked
on Python at that time were well-aware of what was happening
in the functional world, and that this was an important
influence on their thinking about Python. Haskell is mentioned
by name in the following post by Steven D. Majewski:

http://www.python.org/search/hypermail/python-1994q2/0718.html

where Haskell is portrayed as taking one paradigm to
the extreme, as opposed to Python in which all can
be represented. Implying that, in particular, Python
must include features from the Haskell paradigm
as well.

(At that time, I was busy programming in Perl, C++
and other languages, having never heard of either
Python or Haskell.)

Towards the end of the decade, there were more
discussions about adding Haskell features
into Python. Several were implemented starting
in 2000:

"PEP 201 - Lockstep Iteration... Created: 13-Jul-2000... This PEP
proposes a standard way of performing such iterations by introducing a
new builtin function called `zip'...  The function's name...  the BDFL
strongly prefers zip() due to its Haskell[2] heritage."
http://www.python.org/dev/peps/pep-0201/

"What's New in Python 2.0... released on October 16, 2000... The idea
of list comprehensions originally comes from the functional
programming language Haskell...  discussed for a seemingly endless
time on the python-dev mailing list."
http://www.python.org/download/releases/2.0/new-python.htm

Note that after the introduction of list comprehensions -
explicitly from Haskell - Guido's argument against lambdas,
maps, and folds changed a bit. In the blog post you
referenced, his argument is that one should use list
comprehensions instead. So that is not exactly
a rejection of Haskell influence.

Iterators, generators, and itertools we have discussed.

Recent additions to Python include partial function
application (of sorts), type inference (of sorts), and a
"numeric prelude" (of sorts), described in PEPs 309,
3124, and 3141 respectively, all mentioning Haskell
by name.

Regards,
Yitz