4th July 20:00 UTC Monthly Meetup

@paullee I would start at the end result and work your way back.

This is a visualisation of a network graph (it's the building I evolved during my presentation):

..the image itself is generated from a text file by the GraphViz tool with a command like this:

neato -Tsvg -Gpage="4,4" -Gsize="4,4" -Gratio=fill -Gcenter=1 < graph.dot > graph.svg

The input text file used by GraphViz is a good example of how you describe an undirected (possibly cyclic) graph. There are two types of objects defined: nodes, and edges between nodes:

strict graph G {

label="15aee6eef927b01955152d43aa8504d5";

graph [overlap=false];

"0/ll\nOutside" [color="#ddddff",style=filled];

"0/rl\nLiving" [color="#99ff99",style=filled];

"0/lr\nKitchen" [color="#9999ff",style=filled];

"0/rr\nCirculation" [color="#ffff99",style=filled];

"0/rr\nCirculation"--"1/rr\nCirculation" [label="Stair",style=dashed];

"1/rr\nCirculation"--"2/rr\nCirculation" [label="Stair",style=dashed];

"0/rr\nCirculation"--"0/ll\nOutside" [label="2.9"];

"0/rl\nLiving"--"0/lr\nKitchen" [label="4.3"];

"0/ll\nOutside"--"0/lr\nKitchen" [label="2.8"];

"0/rl\nLiving"--"0/rr\nCirculation" [label="4.0"];

"0/ll\nOutside"--"0/c\nStreet" [label="Entry"];

"0/rr\nCirculation"--"0/ll\nOutside" [label="Entry"];

"1/rl\nBedroom" [color="#ff9999",style=filled];

"1/lr\nOutside" [color="#ddddff",style=filled];

"1/rr\nCirculation" [color="#ffff99",style=filled];

"1/ll\nToilet" [color="#ff99ff",style=filled];

"1/rr\nCirculation"--"1/ll\nToilet" [label="2.9"];

"1/rl\nBedroom"--"1/lr\nOutside" [label="4.3"];

"1/rl\nBedroom"--"1/rr\nCirculation" [label="4.0"];

"2/ll\nOutside" [color="#ddddff",style=filled];

"2/rl\nBedroom" [color="#ff9999",style=filled];

"2/rr\nCirculation" [color="#ffff99",style=filled];

"2/ll\nOutside"--"2/rr\nCirculation" [label="2.9"];

"2/rr\nCirculation"--"2/rl\nBedroom" [label="4.0"];

}

It appears that the way you model a network graph in python is very similar. The advantage of using a graphing library is that you can query the graph: check it for cycles or unconnected nodes, find the shortest distance between nodes etc..

Thanks for the detailed explanation and pointer, need more time to study @brunopostle !

It seems 'nodes' in a graph are rooms / spaces; ... 'edges' is the connecting doors at common wall between 2 spaces / rooms ?

Wondering, checking, if it can contain information like ?

which walls / windows define the 'node' (rooms / spaces)
which are common walls of 2 adjacent spaces
'elements' / 'other information' of the 'node' (e.g. furnitures, finishes, area, height, peoples accommodating...etc.)

Similar problems for a geographic maps - finding the common boundaries of 2 countries e.g.

Too many questions :)

@paullee for a building like Villa Savoye, no CAD is going to figure out automatically where one room starts and another finishes.

I don't know how freecad does this, but for IFC I think you are usually expected to manually assign walls as boundaries for spaces. i.e you have to declare that wall A is a boundary for space B, and separately that wall A is a boundary for space C - so the fact that spaces B and C are adjacent is implicit rather than declared directly. I also suspect that when constructing a circulation graph, you would have to check if each wall has a door in order to decide whether to add an edge.

[edit] Thinking about this problem with IFC, a wall can be any length and can span multiple spaces, and there is nothing to indicate which space a door is related-to. So if you wanted to construct a circulation graph, you have to resolve the 3d position of the door, and then make an assumption from this about which spaces it connects.

Thanks again, really have lots to study now !