[cad-linux] Re: CAD with server/client architecture

Thomas,
I agree with you to onehundred percent. I have been a fan of OpenSource
for now more than 8 years and I have seen many ambitious projects die
quickly. I myself am guilty of loosing interest on quite a few over the
years (see e.g. barracuda) whereas my LinAl project is alive and
kicking, allthough the website never gets updated and most of the
activity is behind the scenes. 

What I have learned is, that an approach "design driven by need and
feedback" works well in the OS world. That in combination with small
goals, and small steps. So when I say I am going to experiment with a
parametric solid modelling kernel that feeds from a SQL database, I am
not saying that this is going to be an Acis replacement. Linus didn't
intend to replace Solaris,SCO,Irix,AIX,HP-UX either. Some projects will
pick up momentum over time, some wont. 

The other really important thing is that all succesful OS is always
driven by personal motivation. You know that, so do many many other OS
developers. Once the motivation is gone the project will often die,
except if there are many people with similar motivations. That's why
Linux is alive, so is Apache, SMB, you name it. 

Best regards,
Roland
PS: Sorry to hear about your server trouble. There is just too many
assholes out there.


--- Thomas Schmidt <thomas.schmidt@xxxxxxxxxx> wrote:
> 
> Hi,
> 
> that was me saying :"t's really easy to start coding and get a
> couple=20
> of lines on the screen".
> This thread started with great ideas about client/server design.
> I think that a client/server design can be really helpfull to solve
> lots
> of business requirements. For a living, I work for a telecom company
> designing/developing client/server services. Based on the experiences
> I have made there, I would never choose to implement a CAD system
> as a client/server system as suggested at the beginnings of the
> thread.
> Its not that such a design does not work. Its the amount of work you
> have to put in such a system. The problems are often underestimated.
> Session management, user management, concurrency control,=20
> data integrity, or on a more technical side, protocol definitions and
> implementations, network transparent object references, distributed
> garba=
> ge
> collection, and so on. Building a client/server system from scratch
> is=20
> not for the faint-hearted, even with Corba/RMI/XML-RPC, you name
> it.=20
> 
> But this is also true for many software projects. Many open source
> develo=
> per
> start very excited with something and get bored when it comes down
> to production level quality. Lets take a example, but first let me
> say
> that _everybody_ developing opensource softweare has my respects, no
> matter how far they come. Now consider a OS CAD. The developer
> has already spend some time on this project, and you are already
> seeing some results. Developing is currenty fun and progress is good.
> But there will be a point when the fun vanishs. This is when the
> develope=
> r
> has to implement the _endless_ number of things which are needed for
> some level of production quality, including, for example,
> documentation.
> Some day this CAD program might be really good, but its a long way.
> And this way might be too long for the developer.
> 
> So, whats my point? Start on any project, no matter how ambitious.
> Make a good design which allows to implement feature by feature.
> But do not be suprised that even after two years of development,
> the project is not nearly finished and no longer that much fun.=20
> I would really love to see a CAD program on Linux comparable to=20
> Linux itself, KDE/Gnome, Apache, OOffice,... in their domains.=20
> But its so much work to do, and having lots of abandoned projects=20
> does not help when someone has to decide if his/her company=20
> should switch to Linux.
> Nevertheless, keep on trying :-)
> 
> Bye,=20
> Thomas
> 
> 
> > > The following was supposedly scribed by
> =2E..
> > Remember what someone said earlier in the thread:  something like:
> "it'=
> s
> > really easy to start coding and get a couple of lines on the
> screen"
> > (whoever owns this quote, feel free to jump in and
> > take-credit/correct/flame/whatever).
> >
> =2E..
> >
> > BTW, where is this project? on sourceforge?
> >
> > --Eric
> 
> 

__________________________________
Do you Yahoo!?
The New Yahoo! Search - Faster. Easier. Bingo.
http://search.yahoo.com

Other related posts: