[cad-linux] Re: CAD with server/client architecture
- From: Eric Wilhelm <ewilhelm@xxxxxxxxxxxxx>
- To: cad-linux@xxxxxxxxxxxxx
- Date: Tue, 13 May 2003 00:20:54 -0500
> The following was supposedly scribed by
> Roland Krause
> on Tuesday 13 May 2003 12:03 am:
>So if tomorrow you compile my program on your Windows machine, and I
>have set up a public database that we can both connect to, then if I
>change a few points in that database while you are connected you will
>instantly understand where the advantages of this approach are.
And there is also the drawback. I therefore suggest that the database manager
be able to delegate this task to other database managers (of the same "make
and model" but just playing a different role). If you have multiple teams
(arch, structural, mechanical, electrical, etc (such as in construction))
working on a model simultaneously, each office wants to have the model in a
fairly static state while they are working (at least for a few hours). Major
revisions could have notices sent out, but you don't want your parametric
relationships for the pipe hanger failing while the structural guy tries to
decide where to put a beam or what size to use.
For example: each design team could have a sub-database-manager, which has a
snapshot of the model (which it got from the master-database-manager).
Furthermore, each workstation could have a sub-sub-manager, and each operator
should have the choice of setting the update frequency or whether updates
from others are shown automatically/manually. Parametric modeling programs
aren't really THAT smart (and sometimes the parameters aren't linked
correctly by the operator), so you need to keep things from moving around
while you are working and make your changes in a controlled fashion if
relationships need to be reassigned (ever seen a parametric model twist into
a ball after a control point gets moved?)
I'm thinking that a VFS could be useful here (presenting single tables or
groups of tables as "files" (grouped by editing rights?)), unless slices of
the database could be managed directly by the database server. Maybe the
sub-server and sub-sub-server could even break the groups down into finer
divisions at each VFS along the way (entire model stored on master server,
HVAC model components checked in and out of master and stored on HVAC
engineer's server, AHU, ductwork, etc broken out into groups on that server
and worked on at each workstation involved).
--Eric
Other related posts:
