Why did PUGS die?

[Robert Wilkins]

I'm a programming language designer ( data languages comparable to SQL
and SAS if you want to know ). I'm currently making a decision between
Haskell and Python to implement a compiler where both the source and
target are fairly high level ( think of subordinating SAS data steps
and regression/chisq/etc procs(SAS or S) to the role of assembler and
you get the basic idea ).

So here's the problem: Haskell is supposed to be really good at
compilers , maybe the best, or so I hear: parsing and all the symbolic
transformation that goes after the parse. But it has a really tough
and long learning curve , and methods of thinking that carry over well
( from say , C++ and C++/STL, to Python and lists and dictionaries) do
not translate so easily to Haskell.

By far the most high-profile breakthrough of Haskell in compilers/
interpreters was PUGS, the first compiler for Perl 6. I can't think of
any other well known breakthroughs in that direction. PUGS was a huge
and unique marketing breakthrough for Haskell.

So now PUGS is dead, from what I see on Wikipedia. The developer has
walked away. The Perl 6 community has decided not to get a
replacement, and PUGS has and will have no customer base. Perl 6 has
decided to focus on a non-Haskell implementation. PUGS undoubtedly has
a lot of cool features, components and code. But the PUGS software
product is effectively dead, no maintainer + no customer base = death.

While monads are really cool code transformation stuff, they are also
incredibly tough to grasp, and I am rethinking my preference to
Haskell. And as I've seen over the years, Python is so bloody easy,
especially for someone with a C++ background.

Before I decide, can anyone tell me why Pugs died? The reason for it's
demise would be useful to know whether Haskell is a good choice for
writing compilers. It's clear that it's symbolic transformation and
code transformation stuff is amazing, but it's tough to get.

thanks

Robert

 

[Jochen Lehmeier]

PUGS
Python
Haskell

For me, your question boils down to "should I learn/use Python or Haskell"?

I have no idea about PUGS, but http://pugscode.org seems alive.
http://dev.pugscode.org/timeline?changeset=on;max=50;daysback=90;format=rss
shows changes checked in today. You might just look if they have a mailing
list and ask them directly how happy they are with Haskell.

 

[brian d foy]

Before I decide, can anyone tell me why Pugs died?

It was always a side project and never intended for serious use, and
it's only customers were the developers. No one would really want to go
through all of the work to compile a new GHC to use each updated
release when most of the Perl 6 stuff never worked anyways. Oh, and
it was all very, very slow.

I think you're over-estimating how cool Pugs was. As a proof of concept
and a testbed it was fantastic. For getting anything real done, it was
not all that great.