This Cousin covers the three main mailing lists for the GNU Enterprise project, gnue, gnue-dev and gnue-announce. It also covers the #gnuenterprise IRC channel. A great deal of development discussion for this project goes on in IRC. You can find #gnuenterprise on irc.openprojects.net:6667, or you can review the logs. For more information about the GNU Enterprise project, see their home page at http://www.gnuenterprise.org.
Ryan Grange said that the Supply Chain proposals for GNUe described
ordering parts in for shipment to a customer.
What if you want the part vendor to ship them directly to
your customer (drop shipment) instead of paying freight twice?
Peter Sullivan said that, as there would presumably be
a mark-up to the cost
, then
the way that I would approach
this is to have a virtual "drop shipping" warehouse, and set the business
rules so that any goods "received" into this warehouse were immediately
"issued" to the end customer at the selling price. This would enable you
to calculate your profit/loss on your drop shipping "warehouse" in the
same way as for the other, physical, warehouses.
This was all
a bit theoretical at the moment, however.
John Marciano reported some problems when he
ran the setup.py script
in each directory
but got an error message for
Forms. Derek Neighbors
said that the New
preferred way to install cvs is to run setup-cvs.py
in the root directory of cvs and answer the questions.
You will need to have wxWindows, mxDateTime, Python2.x
installed. Also to do database applications you need
a database and the proper python db driver for that
database.
Also, for GEAS, John got a message
Please don't try to compile
this program, it is obsolete and not maintained any
more - Again this doesn't sound good.
.
Derek explained geas
(the old appserver) is still in cvs because you cant really remove
things from cvs (or in this case we didnt want to) we wanted to leave
for reference. We changed the script so others wouldnt waste time
compiling it. The new application server is in 'appserver'
directory.
John also asked is anyone
using this system in a manufacturing/production
environment?
. Derek said We
are getting ready to roll solutions for two or more small
manufacturing / production shops in the next 6 months.
.
Jeff Childers volunteered to start
fleshing out the SupplyChain.pdf document
. He had
lots of detailed ERP product experience
(from coding to implementation) and tons of opinions :).
Neil Tiffin said he had been in charge,
but am currently under water with my job
so
If you want to start a dialog I would be
more than happy to help.
Ross Sponholtz volunteered to
help out
as well.
Derek Neighbors said Yes we woudl
love it, the source of that pdf is in cvs and is actually
in docbook (xml).
Jeff suggested
Unless anyone has an objection, I think
I would like to use a light UML style to organize the requirements.
That is, define the modules in terms of objects ('actors' or
whatever) with their properties, events and methods.
Neil said he was not opposed to UML, but I
do believe that we will need a business persons view (prose
functionality) of the system design
as well.
Derek agreed - I would like to
see proposals in a narrative form and following the
module guide template (it is of course open to enhancements). I have no
problem including UML along with that, but not as a substitute to it.
This is pretty much because not everyone knows UML or has tools to deal
with it, plus forcing to discuss in narrative clarifies alot.
Jeff agreed, and asked where to find the module guide template.
He asked What application do you guys use to
read/edit the XML DocBook files?
He had tried googling, but this
just found the DocBook specifications, not tools to open it.
Earlier, Neil said that The main stuff
that should be kept or should be built on is in the Base Package.
Previously, the documentation had been built directly
from the object definition and it was not possible for the system
design, documentation, and objects to be different.
However, nobody else seemed to think this was
a big deal and so no need to keep at this
point.
However, he warned that
the work load will only get more onerous as the system grows and
some organization and automation now, while not entirely necessary,
will pay hugely in the near future.
Todd Boyle felt that the "UN/CEFACT
Modeling Methodology"
(UMM)
,
might form a good structure for writing GNUe package proposals.
It is "stackism"
, with
a business process layer
,
a collaboration layer
and
a transaction layer of course.
.
These abstract models were now being turned into
the state machine that defines a
BSI, a business service interface
to keep
the home systems of both/all parties to an interfaced
transaction in sync - That
is the whole entier point of business process software,
that some unambiguous business entity agreed in a
contract, has proceeded from one state, to the next
state or to an exception condition. etc.
Using
UMM, whilst a bit prescriptive, would involve GNUe
in the
middle of a larger community, one that is very clueful and
has huge history and experience, and is throwing off not
only a lot of modeling but pretty specific code objects now,
and... you don't have to write documentation :-)
.
Elsewhere, Alejandro Imass said he was
still around if you need
any help with the original Supply Chain Specs that I
originally wrote and submitted to GNUE a long time
ago.
He personally used
emacs + psgml major mode
to write DocBook - It is by far
THE BEST SGML TOOL IN THE WORLD!
. Neil explained
to Jeff that Alejandro was one of the
original contributors to the specs and while we did not get far
enough to include all of this product structures he had some
very good ideas that were captured in early emails.
He hoped Hopefully with the next version
we can actually build a system.
Jeff replied to
Alejandro If emacs'll run on my Windows
XP box then by george I'll try it.
On Supply Chain,
If you have any notes or ideas that are
still hanging around, even if they aren't pretty or finished, send
them along and I'll review them for the next iteration. Let's
knock the narrative out and then we'll take up the flag again for
a beautiful UML model :)
. Alejandro said that
All sgml tools work like a breeze in
Debian
but In Windoze I would
bet it is much harder to get free ``libre'' (GPL) sgml tools up and
running, but I am sure someone out there has tried and
can help you.
He felt that using a word-processor tool like
OpenOffice was not really ideal for really
large documentation projects
like GNUe.
Further to
so
hows the show this year? more/less people - more/less
interest
? Derek said WAY
more people - similar percent interest - we have a book deal
i think if we want it
. He had been unable to get the
demos working fully so i show VERY
cripled functionality :( - but the marketing slicks are MUCH
improved - so that helps compensate
. He had gone
back to the 0.3.0 official releases rather than CVS for
stability - the cvs demo doesnt
work for designer - m/d is dorked bad in both
.
James noted that master-detail works
perfectly at work
using the
cvs copy
of the code.
The next day, Dan Bethe (dtm) said he had given Derek
a lift back to the airport - i
only got to lwce in time to pack up the GNUe booth! and to fend
off the herds of mad forklifts!
However, he had had a chance
to see gnue for the first time ever!
Further to
does
the name lbergman ring any bells? they submitted our project to
the linux fund
. Later, Peter Sullivan (psu) asked
can someone explain the lburkman
joke/reference to me? I thought it was just me, but jamest
doesn't grok it either
. Nor
do I
, said Jason Cater (jcater) -
I wonder if he signed GNUe up for a
grant, or if he signed up for a grant using the GNUe banner?
there could be a difference :) if it's the first, I'd like to
shake his hand - if it's the second, I'd like to know why -
if it's the third, then someone must be confused because I
only gave 2 reasons
. Peter said
the way dneighbo was talking it
was like it was a known alias - or the name of a prominant
GNUe/free sw hater ;-)
Jason thought there was no
such thing as a GNUe-hater - we're
so huggable and luvable
. James added
we never met a goat we didn't
like
. Peter said he had even
checked to see if it was a name on any of our screenshots, but no
- unless it is the name of the ship on the Navigator screen -
USS lburkman
.
Peter Sullivan said that, just in the same way that
the export, i.e. database -> XML,
bit
of GNUe Integrator was shared functionality
with GNUe Reports, The work that
Jason and Jan have been doing on the on the schema
import wizard
which will
now include data as well as just schema
,
as discussed in
the other
half (import, i.e. XML -> database) of GNUe Integrator, or at
least a major building block for it
.
Derek Neighbors agreed - the
biggest thing that is missing is the data transformation on
the fly stuff. (i.e. Integrator Triggers) and of course a nice
UI to go along with it. I think much of Integrator will
become basically reuse of other core code in different ways within
Desinger.
.
In response to a private e-mail,
Derek Neighbors said that Grid
is implemented to day in xPlatform way simply add rows="3" where 3
represents number of rows in grid to display and you have a grid. Papo
team is submitting scrollbar patch for it. Some day there will be more
traditional eye candy grid, though functionaly its not necessary.
Pop-up forms were already implemented, and
we have some search forms planned as
well
.
James Thompson (jamest) said that the GNU Enterprise main
web server, as referred to in
up and I was hoping we'd just
go to the new site instead of the old one
. Derek Neighbors
(dneighbo) asked
is dcl back
? James said
nope - mail, apache/php, firewall,
user accounts - that's all - lots more to do
.
Peter Sullivan (psu) said he had been
working locally on the new site whilst ash was away - but not
testing as I don;t have Apache/PHP here
- but
with a few hours effort this weekend
we could probably cut over say Sunday to something that is more
up to date than current site - and we can then add the final
flourishes (e.g. polls, if we still want them) over next few
weeks
, suggesting Maybe
we should have a poll on whether to have polls
.
Daniel Prather (MysticOne) asked what
all *does* GNUe do? Right now is it primarily just a contact database?
Daniel Baumann (chillywilly) said it was an ERP
framework - we have forms client, rad designer, navigator, commmon module, etc.
appserver too of course - /me is trying to think if he forgot something
.
Andrew Mitchell (ajmitch) added dcl, bayonne
:)
which Daniel Baumann explained: dcl is a
project management app that is under the gnue umbrella and bayonne is the gnu
telephony server which is affiliated with gnue too - also in the works is an
accounting system using forms (adapted from Nola)
. Jason Cater (jcater)
added also a Rental Management app
.
Daniel Prather asked about point of sale systems. Daniel Baumann, remembering
working on point of sale as we speak -
as I open up a used book store in 6 weeks - and seeing as how I'm a core GNUe
developer, I'm morally obligated to using GNUe :)
. He added
not to point you in another direction, but I looked
at bananahead POS and it
looked pretty darn feature complete - /me didn't actually test it,
though
, adding of course, I encourage you to
use us :) - but if you are needing something immediately, there's a
lead
.
Continuing previous discussions in
that gnue's application framework is not
appropriately generalizable toward groupware
?
or is it that the gnue team isn't interested
in doing so? what if outsiders were to be made interested? are the
models that drastically different between a groupware and a
transactional enterprise framework?
He explained the existing hotbed of groupware
that's got all the development happening, is in the LAMP space
(using Linux, Apache, mySQL and PHP) - and
there's nothing wrong with all that but it's just that the LAMP space
tends to be infested with junior, "not invented here" programming demo
style coders
- so they need an app building
framework which is as easy as gnue - gnue is easy at building specific
kinds of apps - and if it was more fully web enabled (liek with a php or
sashxb style presentation driver) and if it had groupware components, then
someone could use it as a groupware RAD - then we add the ancillary
components such as device drivers, palm sync, and external bridges to WAP
or other messaging and formatting, and b3wm - no more
phpgropuware
.
Later, he asked does gnue have any degree of
collaborative infrastructure? or is it just a concurrently-usable
multiple-individual transaction oriented infrastructure?
Daniel Baumann (chillywilly) said that was what GNUe Common was
for - it has the db absatraction, the rpc
abstraction, and other code fr writing apps (both server and
client)
such as the debug system
,
the xml parser, trigger stuff, the connections
code for linking a form up to a db, the commn object model, format masks,
lots of *stuff* - appserver uses mainly the db and rpc *stuff*
.
Two days later, towards the end of a all night long, off-topic,
discussion, Dan said that the GNUe team
have to not downplay web services :>
.
Daniel said we don't -
the have abstraction layer fr most web service protocols - xmlrcp, soap,
etc.
and appserver should talk
to anything by exposing "web services"
. Dan accepted
that you often want super fast data entry in
a local binary app
but you should have
web access later
. Daniel mentioned the php Forms client written
by Jan Ischebeck. Dan asked what about
generalizing
GNUe as a general web app
platform or collaborative platform/
- he said that Michael Dean
(mdean) had been waving around about that long ago -
but when i talked with derek after lwce, he said that gnue has a
fundmanetally different model (transactoinal oriented or whatever) -
and is not optimized for that directly, fwiw
. He did not see
any real difference, however. Daniel said
you "submit" the form -
you "commit" a transaction - same frelling thing ;)
.
Dan asked is there any reason why we can't make
a *generalized* RAD out of gnue instead of just for database oriented
forms? - i mean you have a drag-n-drop UI for forms and db stuff... why
can't those widgets be, say, user-oriented or task-oriented or
schedule-oriented?
He realised that GNUe Designer was a type of
RAD tool, but only for a specific type of app as
opposed to, say, designing a calendar
. Daniel said that
the appserver will provide objects eventually
- that is very generic - and you don't have to be a db guru then - just
an object lackey
. GNUe Application Server
still needs to use
a db to store
its objects, but I think
objects give you the generic thing that you desire
.
Dan asked are there any alternative RADs which already
exist (like zope or whatever) which are compatible with gnue so we could get
started building groupware on your current gnue shizzit?
Daniel knew of
Zope, but had never used it - we have so much code and
time invested in our framework - I dunno who has the time to adapt to an
existing one - or if it would ever make sense to throw away our stuff rather
than picking some existing appserver
. Dan said that
phpgroupware is state of the art as far as free groupware,
but it's not state of the art in terms of app services
- it
would suck a lot less if groupware developers were
magically granted a kick ass framework to start with
Daniel said
that there had been discussions with the phpGW team - they
do not wish to invest time into
gnue integration
- this had been one of the reasons that
Michael Dean had affiliated DCL with GNUE rather than phpGW.
Dan Bethe said that Dan Kuykendall (seek3r) has slightly
reopened somewhat to the idea of moving toward gnue's direction, one way or
another - but still! that's a bit of a leap
.
He felt that free groupware needed major app
frameworks
that were FSF-compatible, and hence not based on Java -
the only FSF-compatible alternative to the excellent
enterprise quality stuff that's being done in java today, is in LAMP -
which is strong but is not ubiquitous enough - for example it's
web-only
, which could prove unsuitable for tasks
like mass data entry
. GNUe could provide
an alternative, fully free, application framework for groupware
development. He asked if stuff like j2ee be made to sit
atop gnue? or only interoperate via message passing like soap/xmlrpc
?
Daniel said I think all our components/servers
will be web enabled and export remote interfaces
.
Dan said that using GNUe as an alternative to phpGW for groupware
development can't be all that hard on the surface
assuming the guts are ready - assuming that you support groupware-like data
structures - and then simply export it to the web via a standard gnue web output,
like maybe a php display driver similar to the current wxwindows
stuff
.
Jeff Childers asked Has any kind
of data dictionary or data structures definition been started yet?
If not, I have a concept that I've been using over the last couple
years in our development that seems to really shorten development
time.
Properties/fields of objects/tables were defined
as either Primary key
,
Surrogate key
,
Local value
or
Composite value
.
This allowed you to add default logic to
your base classes for data entry controls that determine behavior
when the control is created on a form depending on the data type.
For instance, primary keys would be Never
shown
, whilst surrogate keys would automatically
Implement lookup logic when in edit mode
via right click menu or create a small lookup button next to field on
init etc).
This had several benefits for control and
maintanance, in particular Adding controls to
forms requires only to set the field ID to enable the majority of
functionality that you typically want
-
the control will self-select the
appropriate widget.
Derek Neighbors said
You are describing relevant stuff to
AppServer, however, currently I havent been doing a lot of work in that
area. I think we need to get AppServer farther along before we can get
too into these discussions.
Earlier, Todd Boyle felt that any data
dictionary
should be based on ebXML
and gave the
One sentence elevator pitch: GNUE is not
the only application in the enterprise, and will need to exchange
information with a huge variety of heterogenous applications and
platforms
whether During
the implementation/migration phase at the very least
or
as a permanant interface. He gave some web links for
The
ebXML RegRep specifications
,
Open
source RegRep project
and
The
ebXML core component specification
.
Whilst he appreciated that a
product-specific data dictionary gives you ease of use,
and immediate payoff and reinforces
your own community
, he felt that using ebXML
within GNUe could help bring forward
the day when
individuals and small businesses can conduct exchanges
with each other (as well as large enterprises)
.
EDI had not worked for these groups.
If GNUE doesn't help get us
away from paper invoices, printing and mailing checks,
etc. I'm gonna scream,
.
Christopher Brown felt No,
this puts the cart before the horse
- not all data
would need to be interoperable, and
There _will_ be data on the
system that under NO circumstances should be released externally,
such as security configuration and personal information
(as is found in payroll).
He felt that
The point _isn't_ about
abstract XML schema; it's about the concrete question of what
DBMS tables are on THIS system, and how do they link to the
INTERNAL functionality of GnuE, to forms, business rules, and
such.
.
Jeff looked over Todd's links, and said
Personally, I am 100% in favor of building
an open-standards platform to drive e-transactions. I share your
frustration with the inefficiency of what is largely a 1,000-year-old
process
and agreed that EDI was not the answer. However,
he urged caution - One does wonder, though, with
all the new acronyms these days, just how many standards one project can
support.
Derek Neighbors said that GNUe supported
many standards now. Most of our
project is in XML. We support nearly any SQL92 compliant database
etc, etc.
He emphasised we will
store the data we need. The software is GPL and certainly with Integrator
and our other tools you can get data pumped out in about any format you
want to interoperate with others. Whether it be ebXML or EDI or you
name it.
Jeff agreed - GPL
is as open as it gets. Anyone motivated enough can add a data filter
for whatever standard is required for whichever industry.
Further to
put
in the speaking proposals for the DC conference on "open source and
e-government"
. He noted Even if we
don't get selected to speak, we should submit material for the
"conference handbook" I figure there are about 32 speaking slots (2
sessions each morning and afternoon, 2 days, 4 speakers each
session)
. The conference organisers would
send out notifications of acceptance by Friday,
and we have about a month to get in any material we want to
submit
.
Matt Raykowski (sfb-) asked how
GNUe compares to Compiere.
Peter Sullivan (psu)
said that the big issue for people here
is that Compiere, although free (as in freedom) itself
has non-free dependancies such as Java and (the biggie) Oracle DB
s/w i.e. unless you have sold your soul to Oracle for a database,
it's just a kool but inert lump of code - GNUe is designed to have
no non-free depends at all i.e. you can use GNUe w/Oracle - but you
can also use with MySQL, Freebird, postgresql, SAB-DB etc
.
Matt noted that they're making progress on
PG support though.
. Peter said at
moment, we have tools (very good tools) for building apps - especially
in 2 tier mode (i.e. forms talks direct to database) - but are a bit
short on apps themselves as frankly we really want our appserver working
i.e. would like our official apps to be 3-tier from start
.
Later, Derek Neighbors (deke) said that Compiere had a
qeustionable license
- the
Mozilla Public License - which does not
protect developers
. GNUe was intended to be a full enterprise
resource planning package like SAP R/3,
PeopleSoft and the likes
.
Matt Raykowski (sfb-) asked are the forms
also stored in the appserver and pushed to the client?
Peter Sullivan (psu) said the forms at the
moment contain all the biz logic
but
forms can read forms definitions from an
URL so no big deal
-
once we have appserver working well
(there's already a proof of concept version) then the biz logic sits
there
. He explained geas == old app
server, discontinued - app server = new app server, early in development -
but not everyone is always 100% consistant on terms
. He suggested
trout beatings
for people who got it
wrong.
Derek said on the application side
we are not far along at all - we havent put the focus there - in
short if you want to author an HRMS system gnue could do that today -
if you want to download and 'run' an HRMS system with minimal effort
we can not do that today
. However,
we are strongly encouraging interested
parties in helping make the applications - so we can continue to grow
the tools
.
Matt Raykowski (sfb-) said rumor has it
you're also working with integration with bayonne?
Hans Kugler (LuftHans) said that Rich
of Bayonne was at the GNUe booth at LWE last week.
Peter Sullivan
(psu) said that In theory, bayonne should be able
to use excatly same form defs as full client
, although in practice
phone-based "forms" would need to be simpler.
Hans Kugler (LuftHans) asked what's
wPython.wx and what debian package(s) would hold it?
Jason Cater
(jcater) suggested libwxgtk2.2-python
.
Hans said hmm, looks like I've got the
wxPython modules installed, but gfdes still dies. Do I need to reconfigure
something? rebuild gfdes?
Daniel Baumann (chillywilly) suggested
checking the version of python on the default path. Derek Neighbors
(deke) said that likely wx is either not
installed or installed incorrectly on this machine - we see it all the time
on red hat where people have 1.5.2 and 2.1 or 2.1 and 2.2 installed and
install wx but install it for a python other than the default
.
You could fix this by running setup-cvs.py from a specific version of
python - python2.1 setup-cvs.py - instead
of ./setup-cvs.py
.
Hans asked which util registers
the db connections in connections.conf?
Peter Sullivan (psu)
said vi or emacs ;-) - seriously, I believe
it is just a text file.
Hans had some problems setting up
a connection for a mySQL database. Later, Derek Neighbors (deke)
explained connections.conf gets reloaded
when you run apps
. He could not see any problems with
Hans's setup but i warn you i dont
do mysql so im not much help there - try making the name all lower
case
- iirc i had problems with
this before - fixed it then someone re broked it
.
John Lenton (Chipaca) asked is there
any reason the names of datasources, pages, forms and blocks invade
the root namespace?
He added it
feels ugly, and it breaks things. is there any reason for it?
For example, I can't have a form called the
same as a datasource
and you can
actually name a form in such a way that it conflicts with something
inside gnue - having one entry point would be cleaner, no?
Jason Cater (jcater) said not in our minds -
we have a progressing namespace - as you move up the tree siblings,
etc appear in your namespace. I admit we don't have all the error
checking in place we need - but this progressive namespace was a feature
i.e., a design decision not a bug
. John said
we guessed as much
but
we fail to see the usefulness of it, and we
have several times stumbled over it in dismay - perhaps we fail to see
something you could explain better?
Michael Goebel reported that GNUe
works on my Suse 7.1 distribution with kernel 2.4.0. I tried both
version 0.3.0 and the latest cvs snapshot
. He said
the effort was well worth it. After the
installation was complete (two days), I looked for about half an
hour so at the form samples. After that I had written my
first functional database form in less than 10 minutes. And it
worked !! I already wonder what I will be able to do once I have
actually read the documentation :)
. He outlined in detail
the steps he had needed to install GNUe, which (in summary) were:
I already had a running PostGresql 7.2.1, downloaded from postgresql.org,enabled
for accepting TCP/IP connections
mxDateTime.py from egenix
PyGreSQL (Postgres client interface for Python), resolving some problems with
Making postgres with "configure --with-python".
had to experiment for a whilebut eventually got it to work.
That's it -> test it by running the forms in the sample directory. If in GUI trouble, try passing "--debuglevel=10" as a parameter to gnue-forms. If in database trouble, also take a look at the postgres logfile.