[software] Progress report: ERP systems

[Josh Berkus]

Folks,

Thought I'd give you a progress report.

Because of visiting various OSS conferences, my work to date has 
focussed on evaluating various existing OSS for adaptation to our needs. 
   Since the accounting part of our needs is the most exacting (in terms 
of coding effort) I've been focusing on ERP/Accounting systems.

The major obstacles in adapting an ERP generally have been:
-- Forms: data entry forms are complex and not very modifiable.
-- Contacts: many ERP systems have "customer" and "vendor" tracking 
which doesn't support enough data for donors.
-- Umbrella: "fund accounting" needs support at a fairly low level in 
order to work.
-- Invoice: the invoice-payment-posting cycle is needlessly complex for 
most donations, yet in many ERPs is not bypassable except through a Cash 
GL posting which is really an accountant operation.
-- Reports: Most ERP systems reporting isn't sufficiently adaptable to 
generate the kind of reports we'd need for NPO.  Particularly, most of 
these systems are unable to separate "donated date" from "deposited date".
-- Complex: several ERPs are written in Java or C++ with highly complex 
legacy code bases which would make user-based development and 
maintenance a challenge, and probably mean that this project would never 
be self-perpetuating.

LedgerSMB: lightweight browser-based perl/postgres ERP system. Currently 
not modular or adaptable at all -- creating an NPO version would require 
a fork.  Issues: Forms, Contacts, Umbrella, Invoice, Reports. 
Advantage: is a community-based OSS project with an interest in what we 
want to do.

xTuple: manufacturer's client/server C++/Postgres ERP system.  Currently 
not modular, but supports "custom screens".  Issues: Umbrella, Reports, 
Complex.  Company is supportive but unable to provide any resources to 
our effort.  Also, they are the only ERP I've tried so far that did not 
have a web-based interface.  On the other hand, they have a well-defined 
business logic API via database views, so any use of xTuple would 
probably amount to us creating a brand new web interface on top of their 
database.

OpenBravo: complex J2EE ERP system.  Currently not modular; 
modularization effort is underway and is slated for beta in December. 
Issues: Forms, Contacts, Invoice, Reports, Complex.  OpenBravo company 
is actively interested in our effort, and would provide one staff 
starting in November, plus support from the modularization team. 
However, be warned that the enhanced modularization is probably going to 
mean greater use of Java.  Interestingly, the staff member will still be 
available to help even if we don't use openbravo.

Next up is trying TinyERP and OpenERP.  Also, since I'm back from the 
conferences, I'll want to start a requirements-gathering effort for all 
of the other OSS-NPOs; I know what SPI needs, but other people will be 
slightly different.

Financial software still to be reviewed:
TinyERP
OpenERP
OfBiz

Please add more if you know of something which might be suitable for our 
purposes.

I've also applied for a grant from the Mozilla Foundation.  If I get it, 
I'll be writing up a more detailed list of reference software on the 
FlossFoundations wiki.

Software I'm not planning to review absent doing this for Mozilla 
(because I already know it's unsuitable) includes Compiere/Adempiere, 
CiviCRM, and eBase.


--Josh Berkus