Version française
Home     About     Download     Resources     Contact us    
Browse thread
Re: Why OCaml sucks
[ Home ] [ Index: by date | by threads ]
[ Search: ]

[ Message by date: previous | next ] [ Message in thread: previous | next ] [ Thread: previous | next ]
Date: -- (:)
From: Jon Harrop <jon@f...>
Subject: Re: Why OCaml sucks

Brian Hurt recently published the following blog post "Why OCaml sucks":

  http://enfranchisedmind.com/blog/2008/05/07/why-ocaml-sucks/

I think it is interesting to discuss which aspects of OCaml can be improved 
upon and how but I disagree with some of his points. I'll address each of the 
original points in turn:

1. Lack of Parallelism: Yes, this is already a complete show stopper. 
Exploiting multicores requires a scalable concurrent GC and message passing 
(like JoCaml) is not a substitute. Unfortunately, this is now true of all 
functional languages available for Linux, which is why we have now migrated 
entirely to Windows and F#. I find it particularly ironic that the Haskell 
community keep hyping the multicore capabilities of pure code when the 
rudimentary GC in Haskell's only usable implementation already stopped 
scaling.

2. Printf: I like OCaml's printf. So much, in fact, that I wish it were in 
Pervasives (as it is in F#) so I didn't have to do "open Printf" all the time 
in OCaml. While there are theoretically-elegant functional equivalents they 
all suck in practical terms, primarily due to hideous error messages. I think 
printf is one of the reasons OCaml dominates over languages like Haskell and 
SML. Easy hash tables are another.

3. Lack of multi-file modules: I have never found this to be a problem. Nor do 
I find filenames implying module names to be a problem, as many SML advocates 
seem to believe (yes, both of them ;-).

4. Mutable data: I believe the exact opposite. The ability to drop down to 
mutable data structures for performance without leaving the language is 
essential and the ability to predict memory consumption is essential, both of 
which Haskell lacks. Consequently, Haskell's inability to handle mutation 
efficiently and safely have doomed it to failure for practical applications.

5. Strings: pushing unicode throughout a general purpose language is a 
mistake, IMHO. This is why languages like Java and C# are so slow.

6. Shift-reduce conflicts: although there as aspects of OCaml's syntax that I 
would like to tweak (e.g. adding an optional "end" after a "match" 
or "function" to make them easier to nest), I am not bother about the 
shift-reduce conflicts. Mainstream languages get by with far more serious 
syntactic issues (like <<...>> in C++).

7. Not_found: I like this, and Exit and Invalid_argument. Brian's point that 
the name of this exception does not convey its source is fallacious: that's 
what exception traces are for.

8. Exceptions: I love OCaml's extremely fast exception handling (6x faster 
than C++, 30x faster than Java and 600x faster than C#/F#!). I hate 
the "exceptions are for exceptional circumstances" line promoted by the 
advocates of any language implementation with cripplingly-slow exception 
handlers. I really miss fast exception handling in F#. Brian gives an example 
of exception handling with recursive IO functions failing to be tail 
recursive here and advocates option types. But recursion is the wrong tool 
for the job here and option types are even worse. You should use mutation 
and, failing that, CPS.

9. Deforestation: Brian says "Haskell has introduced a very interesting and 
(to my knowledge) unique layer of optimization, called deforrestation". True, 
of course, but useless theoretical piffle because we know that Haskell is 
slow in practice and prohibitively difficult to optimize to-boot. Deforesting 
is really easy to do by hand.

10. Limited standard library: I agree but this is only an issue because we are 
not able to fix the problem by contributing to the OCaml distribution.

11. Slow lazy: I had never noticed.

The only major gripe that I have with OCaml is lack of a concurrent GC. I 
think this general deficit is going to have a massive adverse impact on the 
whole of Linux and lots of people will migrate to Windows and .NET when they 
see how much faster their code can run.

I have other wish-list items of my own to add:

. JIT compilation for metaprogramming.
. Type specialization.
. Unboxed types (structs).
. No 16Mb limit.
. Inlining.
. Custom per-type functions for comparison, equality and hashing.
. An intermediate representation that I can sell software in to earn a living.
. Pattern matching over lazy values.

I believe these can be fixed by creating a new open source functional language 
for Linux based upon LLVM. However, the lack of a suitable GC is a complete 
show stopper. The JVM is the only thing that comes close and it is unable to 
support tail calls without a catastrophic performance cost, i.e. so bad that 
you might as well write an interpreter.

-- 
Dr Jon D Harrop, Flying Frog Consultancy Ltd.
http://www.ffconsultancy.com/products/?e