[Haskell-cafe] Re: Oracle + Haskell advice?

[Brian Strand]

John Goerzen wrote:
> On 2005-08-18, Brian Strand <bstrand at switchmanagement.com> wrote:
>>I'm thinking about (re)writing some perl code in Haskell (for performance and 
>>correctness reasons).  Has anyone done much with Oracle and Haskell?  So far 
> 
> Not Oracle specifically, but other databases, both free and proprietary.
> 
>>Before I wade in too deep, I'm wondering if anyone has done much "business 
>>app" programming with Haskell (where "business app" is rather fuzzily defined 
>>as database access, web UIs, interfacing with various systems over ip, etc.). 
>>  Does anyone have any advice?  eg "don't go there" or "works great!" or "try 
>>OCaml" :) .
> 
> I'm doing a lot of it.  There's a good deal of web and database activity
> going on here.  Haskell is also my language of choice for scripting
> projects, especially data transfer (thanks in part to my FTP library in
> MissingH <g>)

I'm glad to hear that there are other people (or at least one other person, 
depending on what you meant by "here"...) doing this kind of thing; judging by 
the list traffic you'd think almost no one uses Haskell with a RDBMS.

> I use HSQL for everything database-related.  HaskellDB did not work out
> well.  It proved rather buggy, cumbersome, inflexible (it appeared to be
> impossible to do a "SELECT" without "DISTINCT"), and poorly-scalable
> (try having it scan a database with several thousand tables).
> 
> HSQL is more low-level, but I'm comfortable with that, having used
> Perl's DBI and Python's DB-API before.

This (Perl DBI, ODBC, JDBC) is my background as well; if I can fire SQL at it, 
that's most of the way there.

> HSQL natively supports the common Free databases (MySQL, PostgreSQL,
> SQLite, etc.)  We use PostgreSQL for everything that we can.  It's no
> less speedy than MySQL when correctly tuned, and far more reliable.  It
> also seems to be a heck of a lot easier to install and work with than
> Oracle ;-)

Oracle is indeed a pain to install, although it has gotten better with recent 
versions.  The main things keeping us on Oracle are its very good (online!) 
backup and recovery abilities, high-availability features, and fantastic 
performance tuning (the plethora of wait events and extremely detailed 
statistics is what I'm thinking of here).  Performance-wise it has its strong 
and weak points.

> ...
> You didn't mention what platform you're on, but if you're on Windows,
> I'd imagine you'd have no problem making ODBC work.  Unix is slightly
> more tricky, but again, I'd check out the unixODBC route.  It's included
> in Debian by default, and I wouldn't be surprised if other OSs/distros
> have it too.

We run Suse 9.3 on x86 and x86-64; unixODBC does come out of the box here.  If 
Takusen doesn't work out for some reason, I'll check into HSQL + unixODBC.

Thanks for the advice,
Brian