Re: GNU Cash (lack of) support of Postgres

bruno(at)wolff(dot)to (Bruno Wolff III) writes:
> I ran accross the following in the preliminary FC5 release notes:
> The PostgreSQL backend for GnuCash has been removed, as it is
> unmaintained upstream, does not support the full set of GnuCash
> features, and can lead to crashes. Users who use the PostgreSQL
> backend should load their data and save it as an XML file before
> upgrading GnuCash.
>
> This isn't the kind of thing that will help spur Postgres adoption.
> I don't use GNU Cash and so don't have much incentive to followup on
> this, but I thought there might be someone in the Advocacy group
> that might want to talk to the GNU Cash guys about this.

I used to be involved with the project; haven't had time in quite a
while, which prevents me from being terribly influential about the
matter.

What they are thinking of doing is to adopt SQLite
<http://www.sqlite.org/> as a storage system.

Basically, they're NOT interested in pushing any burden of database
administration on users, which means they aren't keen on:

a) Requiring particular additional packages to be installed

(there is no lack of dependancies already)

b) Requiring management of pg_hba.conf

The users *they're* keen on getting are ones that want to get
something to replace Quicken, to whom configuring a DBMS would
seem like overkill.

By the same token, they immediately *lose* several things, by ruling
out PostgreSQL in favor of pretty well anything else...

- PostgreSQL has Good Numeric Types for Money, and very robust
data types in general
- Stored procs could be very helpful for balance analysis
- There are places where they could *really* use triggers

But the two big reasons are ones they are evidently allowing to
override things.

Forcing PostgreSQL on people would be something of a non-starter,
thereby making the would-be benefits pretty irrelevant, and when they
are having a fairly tough time time with the existing burden of code
(which they need to port to GNOME 2, for instance), adding effort,
with no actual benefit (because they can't trust the database with
types, procs, or triggers since they won't use it universally).
--
let name="cbbrowne" and tld="acm.org" in name ^ "@" ^ tld;;
http://www.ntlug.org/~cbbrowne/nonrdbms.html
"Those who doubt the importance of a convenient notation should try
writing a LISP interpreter in COBOL or doing long division with Roman
numerals." -- Hal Fulton