[Haskell-cafe] Re: Web server (Was: Basic question concerning data constructors)

[Joost Behrends]

Thanks for that info:
> 
> Several people have adapted and further developed HWS:
>  
http://www.haskell.org/haskellwiki/Applications_and_libraries/Web_programming#Web_servers
> 
>   http://darcs.haskell.org/hws/
>   http://www.informatik.uni-freiburg.de/~thiemann/WASH/ (WSP)
> 
> There is also a mailing list dedicated to Haskell and Web development:
>   http://www.haskell.org/mailman/listinfo/web-devel
> 

I've already browsed through the docomentation of all that. Sorry, but i will 
not use WASH. I like things to be direct, to write >> p { ... } or similar
things instead of <p> ... </p> is worsening things for me.

Same with SQL.

To work competently with CSS or SQL you must learn that languages anyway.
(I still didn't master CSS2.1, but will have mastered it in the near future.
Haskell's pattern matching has helped me much to understand the selectors of CSS
- but problems with the "natural flow" of elements and positioning
are still there).

Using WASH (or HaskellDB) for me only means, that i must learn another 
interface on top of that. For databases i could imagine to do that for

1. Real database abstraction (which would require far more complete drivers, 
ODBC for example)
2. Mighty ORM abilities. 

With anything less, i do not want that extra learning. I consider the most
approaches as megalomanical, which
try to "improve" SQL. SQL is an ingenious achievement with its complete
avoidancs of iteration - it's working with sets - in an area, where performance
is everything. Still now it's a model for many things in language design -
i remember to have read something corresponding on www.haskell.org even (don't
remember the context).

For CSS (i try to do most things there, not in HTML) i see no use for that
extra learning at all.

Please do not feel offended. Sometimes i am too rash with hard words.

And concerning the web servers: I haven't seen any indication, that someone
has ported hws to Windows - as easy as that is. There is not even use of
unix domain sockets in hws, nothing unix-specific with IPC (or communication
among threads). It's just EpochTime and access permissions and only changes
in Utils.hs and Main.hs have to be made it seems (The other modules compile
on my system at least). If someone had done that, it would be in the core code 
of hws.

However it is kind of nasty for a Haskell beginner. The Windows API is
accessible via FFI, but nearly undocumented (and i am used to portability, do 
not want to study it earnestly. We have the module os in Python - i am missing
that in Haskell).

And time looks complicated in the standard lib. It was even an obstacle for
compiling HDBC (old-time had to be exposed - that is, what i thought at first,
but had to find out then, that cabal ignores exposition and hiding of modules 
:[[  ).

Cheers, Joost