gavino

happs
http://happs.org/HAppS/README.html
Prospective Good News

Haskell is more expressive than other languages so you can create DSLs
for your application domain and define business logic more efficiently.


Having the DBMS as part of the app means you don't have to write all
the database marshalling/unmarshalling code typical of LAMP. So you can
develop much faster.

The type system means that it is difficult or impossible to make a
large class of errors that are incredibly typical of traditional
database programming. So you are less likely to run into costly run
time errors.

Integrating the pieces together in memory means you are not expending
the huge amounts of CPU time used in LAMPs marshalling data between the
HTTP server and the app server and then the app server and the DBMS and
the underlying disk. So you can save money on buying, maintaining, and
hosting large numbers of CPUs.

Integrating the pieces together also means that you deploy a single
integrated and tested executable and are not as much at the mercy of
the vagaries of the deployment environment. So you can save money on
sysadmins.
What applications benefit from HAppS?

HTTP requests and SMTP envelopes encapsulate transactions and not vice
versa.
Note: doing otherwise with LAMP is considered bad design because it
implies a requirement to maintain and garbage collect database
connections arbitrarily. So this should not be a high hurdle.

All operating data fits in memory (store blobs on disk.)
Note: Although this seems like a high hurdle, COTS servers with 12gb of
memory are readily accessible and some vendors let you reach up to
500gb of RAM. FYI, eBay has around 50M active users. If you maintained
1k of queryable data for each of their users, you would need only 50GB.
(You would also need to recompile your app for 64bits so the math is a
little more involved but you get my point).

You don't need more CPU power to server your app than you can obtain in
a single machine.
Note: I have not benchmarked this code yet, but another Haskell server
was benchmarked at near 1000 HTTP transactions per second on a Pentium
4 in 2000. Modern web servers with similar architecture can serve 10k
HTTP transactions per second. eBay serves 400M page views per day,
which comes to an average load of 5000 hps and a peak load of perhaps
50k hps. In other words, an OTS 8 CPUs system, could handle all of
eBay's HTTP traffic

ari johnson

"gavino" <bootiack@yahoo.com> writes:

Please stop it with pasting the same off-topic crap. What feature do
you want? If it exists elsewhere, why are you bugging c.l.l about it
so much?

vagif verdi

With the same amount of work that was required to write happs in
haskell.
)

Will anyone do it for you ? Sure, why not. Are you willing to pay for
the work ?