Re: Haskell Servlets (was Re: Questions from a returning Haskelluser...)

[S. Alexander Jacobson]

On Tue, 4 May 1999, Sarah Thompson wrote:

> >I'd hate to think that anyone might misinterpret Alex's comments here.
> >Haskell is not limited to the Win32 platform, and neither are the tools
> >that I think Alex was refering to.  Both Hugs and GHC can be used on a
> >range of different platforms, including linux and unix.  Things have
> >been this way for a long time and, to the best of my knowledge, everyone
> >involved with these systems is committed to continuing cross-platform
> >support and development for the foreseeable future.

I did not mean to imply that there was anything about Haskell or GHC that
was platform specific.  My comments were focused on library issues and
ActiveHaskell.  My complaint is that the work done there, though
exciting, might have been better applied to getting Haskell talking to
Java and using Java's copious libraries.  On win32 platforms, Haskell
could then exploit Java's extremely clean COM support.

In simpler terms, I have developers on linux, solaries, and win32
platforms and I don't want to force them all to run win32 just to use
libraries.

> Our choice of NT rather than some other Unix was simple expedience - we
> had plenty of experience of it, machines running it, development software
> for it and so on, but most importantly we knew that there were people out
> there willing to *pay us* to write software under NT. Going the Unix or
> Linux route would have been financially suicidal - a small software house
> that is self-funding development off the back of development contracts can
> not afford too much altruism.

I did not intend to criticize Sarah's choice of NT for her deployment. If
it makes sense for her business that is enough for me.  My point here is
to make sure that we don't end up going in directions that lock us in
towin32 e.g. starting with COM and then following a natural path to ADO/ODBC
rather than JDBC.

> However, we did go to considerable lengths to
> encapsulate as many NT-isms as possible in C++ classes with a view to making
> a later port more straightforward. Now that threads are available and well
> supported under Linux, a port is at least now feasible without a total
> rewrite.
[...]
> We have taken the decision that the next major version number update of Telergy
> will support Linux and NT, as well as (possibly) a number of Unix variants.

Excellent.

> I have to be a CEO first, because that is
> my job. However, I have believed strongly for quite a few years that
> functional languages offer extraordinary potential benefits - I will be very
> pleased if by releasing a Haskell-based web interface to Telergy that more
> developers will get their chance at a first taste of this new world.

Being a CEO myself, I completely understand your position.  We are in fact
using Haskell in production now and would like to extend that use much
further (instead of Perl/Java).   We also have extremely limited
time/resources to devote.  But, from your initial description, it
sounded like you would be covering some of the same ground that we had
covered in exploring the issue before.  Moreover it appears that Jan has
also covered some of that ground.  My suggestion is that we collaborate
in making this happen.  By participating you are trading  application
specificity to your project for the combined expertise of people on this
list in making and supporting a working Haskell application server
framework.  

Because I assume this discussion is not of direct interest to everyone on
this list, I just created a separate discussion list on egroups focused
exclusively on building such an application server.  You will recieve an
invitation to join shortly.  I do hope you participate.

I will reply to Jan's posting there.

-Alex-

___________________________________________________________________
S. Alexander Jacobson                   Shop.Com
1-212-697-0184 voice                    The Easiest Way To Shop