Working on my network lib, I got into some other pb/limitations :
The way I'm building the lib is the following :
- build the basic objects (basicaly an rj45 model and a fibre model)
- include them in the lib
- use those objects to build more complex ones (row of 4,6,8 RJ45,
tranceiver)
- include them in the lib
- build even more complex object using the preceding ones (mainly
daughter cards to put in some chassis)
- include them blablalbla
- finaly build your hubs, switches, router ...
I already did all that a year and an half using xfig and it seems to
work.
BTW is the rotation of an object already in the CVS version ?? I had to
build both the row and the columns of RJ45 by hand. Feel it a bit boring
;-)
Now, I had that strange idea to but a connection point in the middle of
my RJ45 object just in case I want to connect it to a
station/switch/router etc ...
Build my row of 4 rj's. Saved it, exported it to svg, build the custom
shape and lost the information about my connexion points :-((
I had a look at the svg file nothing in there (normal), had a look at
the .dia file nothing in there either, about the position of connexion
points (sure, everything is in the custom shape and in the dia file you
just got some position and properties stuff).
So I build all my rows and columns of rj's adding the connexion points
by hand (with the right initial position and the right scaling, it's not
that much painful).
Now I start building the switches ... and that really sucks. With rows
of rj's on tranceivers or on daughter boards which are themselves on
chassis or switches it's almost impossible to find a scale/position
where the position of the connection points will be easy to find. And
now I've 16, 24 position to find (I've also a 80 ports switches in my
todo list). And as evry body nows, hand work doesn't scale well ;-))
So if so got a nice idea of how not to lose the information about the
connexion points through the .dia/.svg/.shape cycle... I'm open to any
advice :)
Now for those who have read this far, we've tranceiver/daughter cards on
my left and chassis on my right. In the middle a user, who wants to
build his/her switch/router using those components ...
The first way of doing it is just to put the element you want on the
switch and zooming in and out to get it right. That's what I used with
xfig but that are a lot of pbs with that approche :
- difficulties to get your objet perfect (alignement in all directions,
right position for each module you put etc ...)
- there are small chances to be able to build exactly the same object a
second time
- trouble with scaling/moving
So I had a look at visio this week and they solve that problem nicely
using handle and connexion points :
- they use connexion points on the chassis to mark the position of the
tranceiver
- they put handles on the tranceiver
- and the user has just to clip his module on the right place to get it
be a part of the chassis.
Is there a way to implement that fonction in the custom objects ??
BWT, in the README at the root of the CVS, there is the following on
handles :
"
Note on handles and connection points:
--------------------------------------
An object has handles to resize it. A handle can be moved either because
the user dragged it with the mouse, or the handle is attached to another
object, which moved itself. The handles are diplayed as little squares
(red: normal, green: attached to an object, blue: can't be moved).
"
Isn't there a mixup between green and red ;-)
Good thinking,
Fab
