[HACKERS] [RFC] Unsigned integer support.

Hello hackers,

I know the development community is in the middle of the July 2008
commit-fest, so I apologize if this design proposals are in
appropriate at this time.

I am looking to take advantage of PostgreSQL extensible type system
and implement unsigned integer support. The data I am dealing with
contains
many unsigned data types and I am planning on using this type to
reduce storage consumption. I am initially looking to add support for
16-bit and
32-bit unsigned integers with the potential to add 8-bit and 64-bit
unsigned integers if needed or desired by the community.

Searching through the list archives, I found two objections raised in the past:

1. Implicit casts between different data types.

I am hoping the removal of many of the implicit casts in
PostgreSQL 8.3 will simplify this task to where this objection can be
removed.

My plan (without much experimentation) is to have maybe a handful
of casts (something like):
* uint4 -> integer
* integer -> uint4
* bigint -> uint4
* integer -> uint2
* uint2 -> smallint

and then provide operators to provide a reasonable set of
functionality. My initial thought for this functionality is to
provide default operators on any
type that is not implicitly casted on the psql command-line.

As an example, I am planning for the following SQL statements to
work correctly:

3000000000::uint4 + 10 and
3000000000::uint4 > 10

My understanding is the SQL standard does not provide support for
unsigned integers, so I am planning on making all casts from unsigned
integers
to other data types explicit. Is this acceptable to the community?

Another question for the community is should we allow the following cast?
-1::uint4

Even though this is acceptable c-code, I am leaning towards
throwing an out-of-range error when this occurs.

Are there some areas I am missing or should investigate further
before working on this project?

2. There is not much demand for unsigned integer types.

Not much I can do about that :) I am willing to post my work as
a PgFoundry project.

PgFoundry already has an uint project:
http://pgfoundry.org/projects/uint/

Unfortunately this project seems to have not gone anywhere. Last
activity was late 2006 and there are not any files checked into the
SCM repository.
Is it acceptable to hijack this PgFoundry project? Or should I
start a new project (assuming there is any interest in publishing this
work).

Although I am not targeting inclusion for this type in the core
PostgreSQL code, I would like to post code for review and receive
feedback from the
community on this work. As I understand this RFC is the first step in
the process :) Once I have some code ready for review, is it
acceptable to use the
commit-fest wiki for this project?

Thanks much for your time!

- Ryan

--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers