[Gnumed-devel] [GPCG_TALK] Gnumed (was: GP split on software ads)

[Ian Haywood]

Sebastian Hilbert wrote:
> 1.) so called shared finance modell
> - people commit to paying a certain amount one a predefined amount (total) has 
> been reached
The two lists I subscribe to are gnumed-devel and an Australian GPs list
called GPCG_TALK

Currently a thread is running around this same issue, I was drafting an e-mail 
to unite them but Tim beat me to the punch.

Tim Churches wrote:
> However, there is still a HUGE amount of work to be done before GnuMed
> is ready for production use. The GnuMed team is still wrestling with
> (and iterating over) the design of fundamental aspects of the programme.
> Version 0.1, with a full working GUI, is yet to be released.
This is correct. All of us are active clinicians and this limits development 
speed very greatly,
My wife once asked me "when will you ever finish this Gnumed thing", I said,
"It's like trying to build a ship in your spare time, as you're trying to weld 
on the propeller,
a bunch of guys bang on the shipyard door and demand to know when they can start 
sailing..."

IMHO, Horst's inability (through no fault of his own) to play his earlier 
leadership role has been
the biggest reason why we have been unable to deliver a basic working system 
over the past 2 years.

> (If anything, I feel that GnuMed is in danger
> of becoming over-designed and over-engineered, and that the absolutely
> excellent has become the enemy of the good enough, but that's just my
> opinion).
This is fair criticism to some extent.
One of the problems we have is Gnumed is not like most other FOSS projects.
Most other projects their design documntation is "we are writing a free equivalent 
of X"
where X is UNIX kernel, Microsoft Word, PGP, etc.
If Gnumed was "we are writing a free equivalent of Medical Director", we would 
already
be finished. However, MDW isn't even "good enough", so we become bogged down in 
design issues, we make a system
that is configurable to the nth degree, etc., but in the long run I think this 
extra effort is worth it.

> My gut feeling is that full-time work by a team of three
> or four people for 4 to 6 months might do it (plus continuing input from
> volunteers). A full-time team is important,
There are professional "Python shops" here in Australia.
However, most professionals would, I suspect, want a defined project
of their own, I'm not sure you could give someone a paycheque and a CVS password
and say "get to work with want we've got".

Once we have a basic demonstrable system (in Australia this means notes, path. 
requests, path. results, plus simple prescribing)
with defined structure, documented internal APIs, etc, it becomes more 
interesting. Isolated mini-projects can be generated
(such as decision support) which can be contracted and funded by external 
parties.
Porting to e-smith and generally making installation painless is such a 
project, too.
Personally, I am keen to do an RACGP research job somehow involving gnumed, 
but, again, I need somehow a defined 6-month project.

It's getting over this quantum barrier that's proving hard for us alone,
and difficult for external people to help with.
Starting from scratch with contracted programmers is an option, naturally
I have a vested interest against this as so much of our work will be wasted.

One option is the precedent of Larry Wall (the developer of Perl)
He solicited donations, promising once the total reached, as I recall, 
$US45,000,
he would quit his day-job and work 100% on Perl for 12 months, which he did.
However I'm not sure which, if any, of the Gnumed developers would be prepared 
to
do this (or, more importantly, how many would be prepared to donate!)

> One other possibility is the forthcoming Vista Office suite.
As far as I can tell, VistA is closed-source and Windows-only on the
client, with the client-server protocol kept under wraps
for "security reasons", can anyone clarify this?
Otheriwse, VistA is very interesting.

Ian Haywood

signature.asc
Description: OpenPGP digital signature