Wayne Wilson wrote:
> Thomas Beale wrote:
> >
> > In our judgement, the use of good formalisms and tools makes for a significantly
> > better result, and does not unduly harm the economics of development (since
> > development has to be funded from somewhere - nothing is really free, and the
> > cost of wages dwarfs the cost of tools), so we prefer the "quality" approach,
> > even if it costs something.
> >
> I am sympathetic to your issue here, but open source makes another
> argument about quality and robustness which is that the more people who
> have access to the source code, can examine it, make changes in it, put
> it out to test, the higher the quality of the end product.
Agree. But consider this: the more specialised a product/project, the fewer people who
will work on it. I think it is reasonable to say that far fewer people than work on
Linux will work on (e.g.) the GEHR kernel. And for most people on this group, there
will be no need to compile the GEHR kernel - if they are building applications, they
just want to link to it in a binary fashion (e.g. via .lib, COM etc). The source will
still be available of coruse - no problem there. But I think it is worth always
considering the actual facts of the general argument here. How many people does it
really affect?
And the other side of the coin is that if we are only allowed to use e.g. C++ or Java,
we have compromised the on the eventual quality of the thing. Believe me, the
deliverables of an EHR project - requirements, architecture/model, implementation are
nothing like the Linux kernel. The thing is requirements driven, and information
centric, whereas Linux (unix) is (still) largely task/function-centric. For us, the
compromises forced by C++ (cost of development due to complexity) and Java (semantics
too weak) were too big for the GEHR model and kernel. But I have no problem whatever
with people developing Java applications which talk to the kernel. In fact, I think it
is a good thing to do - it uses the strengths of the language in the right place.
In any case, I have a feeling that the kernel itself will compile under Small Eiffel -
it is the Matise binding which will not (but whcih can be supplied as a binary .lib).
So I don't think the situation is all that difficult. If was important for people to
be recompiling the source of the GEHR kernel all the time, I would try to ensure that
it could be done under SmallEiffel.
> The collective cost of the community contributed wages dwarfs any one
> organizations wage costs.
>
> It has turned out that having a very powerful C,C++,Java environment
> for free running on other free environments (gcc and linux for example)
> creates such a scenario. That this environment is not what we would like
> means that if we are to have open source projects thriving using our
> preferred tools, we must make a similar effort to get those tools freely
> available to everyone.
As I say, not "everyone", just the interested parties. It will be a much smaller
number.
> While $750/developer may seem like a pittance to those of us
> professionally employed in the US, it may seem like a mountain to a
> volunteer in another country. There is another subtle sub-text here that
> is built in to the high cost western way of doing business, that if you
> can't keep up with our expenditures, you can't play in our
> neighborhood. I don't think anyone is conscious of this message so I am
> not accusing anyone of deliberate intent.
Yes, I understand this problem, and sympathise. (Actually I do other work in community
informatics for free or on a grant basis, with developing world and academic people
in mind; I am quite aware of the problems...).
I feel that time is of the essence, and that we should get something of good quality
built, which most people can link into their application, which may be built with free
Java tools or whatever. If developments over time lead us toward more people modifying
the GEHR kernel or the Matisse work I have done, then it will clearly be worth making
efforts to ensure those people are not penalised by tools costs.
- thomas beale
--
---------------------------------------------
Deep Thought Informatics Pty Ltd
Information and Knowledge Systems Engineering
mailto:[EMAIL PROTECTED]
http://www.gehr.org
phone: +61 7 5439 9405
---------------------------------------------
