Small dedicated install of PgSQL for a program

Hey all,

I've got a program that uses PostgreSQL. In the past, one of the
trickier parts of installation and design was supporting various
versions and various types of SQL servers. So now that I am doing a
ground-up rewrite of the program, I wanted to use a dedicated
installation of PgSQL.

I've got it compiling and installing nicely, but I am looking to
"optimize" things some. Is there any guides for people like me looking
for a minimal install/distribution size? I'd like to strip out all
features I don't use. Also, I'd like to avoid needing the user to
install <foo>-dev packages, so is it possible to get a list of all the
library and include files PostgreSQL needs?

So far, I've got this working;

./configure --prefix=/usr/share/mizu-bu/db/ --without-docdir \
--with-perl --with-pgport=1062 --without-tcl --without-python \
--without-krb5 --without-pam --without-ldap --without-bonjour \
--without-openssl && make && make install

The '/usr/share/mizu-bu' directory is the root directory for my
program and PlPerl is the only language I need. I don't need a lot of
the stuff in contrib (ie: tsearch2 for example), but am unsure of how
safe it is to just start deleting things.

Hopefully others have created (relatively) small, stripped down
distribution of PgSQL for their programs...

Also, to make a source distribution as portable as possible, what
precautions/gotcha's might I run into by trying to put together a fully
independent list of libraries and includes for this install? I'm far
from an expert programmer, so I am not sure what all I need to watch
for. I know I need GNU make 3.81.1+, and I'd like to have readline
support (50/50 of zlib).

Thanks for any help/tips/pointers!

Madi