On Mon, Oct 25, 2004 at 09:28:23PM +0200, Tomasz Zielonka wrote:

On Mon, Oct 25, 2004 at 08:55:46PM +0200, Remi Turk wrote:

...

P.S. Why do so many people (including me) seem to come to Haskell from Python? It can't be just the indentation, can it? ;)

How many? I don't.

Best regards, Tom

At least one. (Me) And, judging from the amount of references to Python in these mailing-lists, I really doubt I'm the only one. I actually met Haskell mostly by reading about it in the python mailinglist/newsgroup. (in e.g. Alex Martelli's posts) Someone else (who shall remain anonymous. SCNR) wrote:

Speculation I'm afraid to post: People are drawn to Python because they hear it is a clean language, but slowly find that it's really pretty messy internally. Haskell is beautiful on the outside and the inside. :-)

The last sentence I can definitely agree with, but I'm not so sure Python really is that messy. (Messier-than-Haskell sure, but messy? ;o) Groetjes, Remi P.S. Hm, it _is_ haskell-cafe, but maybe it's about time for a [Off-topic] note? ;) -- Nobody can be exactly like me. Even I have trouble doing it.

Shae Matijs Erisson wrote:

Remi Turk writes:

...

At least one. (Me) And, judging from the amount of references to Python in these mailing-lists, I really doubt I'm the only one.

At least two. I also came to Haskell from Python.

Me three.

...

I actually met Haskell mostly by reading about it in the python mailinglist/newsgroup. (in e.g. Alex Martelli's posts)

I met Haskell because I started writing all of my Python code with a single return point, was overusing reduce and map, building function pipelines, and finally someone asked me if I'd used Haskell before.

I met Haskell by first meeting Vyper, which was a more functionally-slanted variant of Python, which was written in Ocaml. This led me to Haskell, which I liked better.

...

...

Haskell is beautiful on the outside and the inside. :-)

Yes, Haskell is beautiful inside and outside. That's something that has kept me interested in it for years. 'Conceptually pure' is what I call it.

Yep. Bryn

Tomasz Zielonka writes:

...

P.S. Why do so many people (including me) seem to come to Haskell from Python? It can't be just the indentation, can it? ;)

How many? I don't.

And I don't either. But indeed I've seen more references to Haskell on Python lists than on other lists: http://groups.google.com/groups?&q=group%3Acomp.lang.python+haskell - 1590 posts http://groups.google.com/groups?&q=group%3Acomp.lang.python - 219 000 threads http://groups.google.com/groups?&q=group%3Acomp.lang.java+haskell - 73 posts http://groups.google.com/groups?&q=group%3Acomp.lang.java - 85 100 threads http://groups.google.com/groups?&q=group%3Acomp.lang.perl+haskell - 35 posts http://groups.google.com/groups?&q=group%3Acomp.lang.perl - 52 700 threads http://groups.google.com/groups?&q=group%3Acomp.lang.c+haskell - 197 posts http://groups.google.com/groups?&q=group%3Acomp.lang.c - 435 000 threads http://groups.google.com/groups?&q=group%3Acomp.lang.scheme+haskell - 1330 posts http://groups.google.com/groups?&q=group%3Acomp.lang.scheme - 44 800 threads I don't know how to get the total number of posts from a group archived by Google. Assuming that average threads on these groups have the same size, Haskell is - 8 times more popular on c.l.python than c.l.java, - 11 times more popular on c.l.python than c.l.perl, - 16 times more popular on c.l.python than c.l.c, but finally - 4 times *less* popular on c.l.python than c.l.scheme, i.e. the order of Haskell-awareness is Scheme > Python > Java > Perl > C. -- __("< Marcin Kowalczyk \__/ qrczak@knm.org.pl ^^ http://qrnik.knm.org.pl/~qrczak/

On 2004-10-25, Tomasz Zielonka wrote:

On Mon, Oct 25, 2004 at 08:55:46PM +0200, Remi Turk wrote:

...

P.S. Why do so many people (including me) seem to come to Haskell from Python? It can't be just the indentation, can it? ;)

How many? I don't.

I'm one. I look at Haskell at what Python should be in a lot of ways. Haskell's type system is static and helpful, yet inobtrusive like I like my Python. Python has added a lot of FP-like features lately, and they're great. Things like generators, etc. But why go only half-way? Haskell doesn't need a special generator because a list is lazy. I like that. Haskell's greatest deficiency compared to Python, IMHO, is the breadth of good libraries available for it. I'm working to fix that. One down (ftplib), many more (ConfigParser, etc) to go :-) -- John Goerzen Author, Foundations of Python Network Programming http://www.amazon.com/exec/obidos/tg/detail/-/1590593715

On Tue, Oct 26, 2004 at 10:22:23PM +0200, Tomasz Zielonka wrote:

My road to Haskell went from C, C++ and Perl through Ocaml, Clean and Erlang. I was mainly motivated by dissatisfaction with languages I used at the moment. I am very happy with Haskell and I think I'll be using it for some time. I still use C++ at work, mostly because of performance requirements and participation in multi-developer projects (multi stands for 2-3 here). To be honest, I am very happy with what we managed to achieve in C++, but I still get angry when I code some more complicated but less-performance-critical parts and I think how easy it would be in Haskell :)

All C and assembly programming for kernels and firmware at work here. -- wli

Marcin 'Qrczak' Kowalczyk writes:

I would not be surprised if relying on GC to close open files would be generally considered kosher in a few years - in cases when it has little visible effects outside, i.e. excluding network connections, but including reading configuration files.

One of the goals of ReiserFS, if I recall correctly, is to create a filesystem API that doesn't involve file handles. I wonder how far one could get without explicitly dealing with file handles outside of the IO library implementation. -- David Menendez http://www.eyrie.org/~zednenem/

David Menendez wrote:

One of the goals of ReiserFS, if I recall correctly, is to create a filesystem API that doesn't involve file handles. I wonder how far one could get without explicitly dealing with file handles outside of the IO library implementation.

I tried to do this last year with my File and Directory abstractions [1], but the consensus was that it would be impossible to implement on top of Posix or Win32. It might be possible on top of the native NT API, though. For this to really work properly the whole OS needs to be designed around it. Such OSes exist -- they're called "capability-based" -- but like so many other good ideas this one hasn't generated much interest in industry. I think we're stuck with explicit file handle management for the next couple of decades. -- Ben [1] http://www.mail-archive.com/haskell@haskell.org/msg13071.html

Ben Rudiak-Gould writes:

...

For this to really work properly the whole OS needs to be designed around it. Such OSes exist -- they're called "capability-based" -- but like so many other good ideas this one hasn't generated much interest in industry. I think we're stuck with explicit file handle management for the next couple of decades.

On Tue, Oct 26, 2004 at 09:22:58PM +0200, Shae Matijs Erisson wrote:

hOp/House doesn't have a filesystem yet. See http://www.cse.ogi.edu/~hallgren/House/ (source recently released) Maybe we won't always require explicit file handles.

It doesn't seem to have demand paging either. Hmm. Maybe I should look at this sometime. -- wli

Sam Mason writes:

The best idea I've seen is another one from Microsoft Research. It's an extension to C that allows the the programmer to use the type system to specify the lifetime of things.

I'm worried about putting too many thing in types. For example many people believe exception specifications in Java are a mistake. Such problems often don't show up until someone tries to write big programs. It's not sufficient that it appears to work for toy programs. Putting many things in types is generally problematic for all kinds of polymorphism. -- __("< Marcin Kowalczyk \__/ qrczak@knm.org.pl ^^ http://qrnik.knm.org.pl/~qrczak/

On Tue, 26 Oct 2004, Graham Klyne wrote:

At 20:55 25/10/04 +0200, Remi Turk wrote:

...

P.S. Why do so many people (including me) seem to come to Haskell from Python? It can't be just the indentation, can it? ;)

I did. (Or: from Java via Python.)

I don't think it's the indentation. At least, not *just* that.

Maybe it is the fact that Python adapted some of the techniques that make Haskell convenient, like map, filter, list construction, lambda and so on. Perhaps people like to know the origin of these extensions?