# The following was supposedly scribed by # Chris Somerlot # on Wednesday 25 August 2004 11:05 am: >> No, the geometry is segregated from the header entities. =A0Whether they= are >> in the toplevel as "<ID>.yml" or not is another issue. >> >> I suggested that the geometry is actually in a "model/" directory under >> the toplevel, but Bruno and I didn't exactly come to an agreement on that >> one. > >I'm confused about how we got away from the flat directory method.=20 We haven't really. The importance of the flatness is only in assigning=20 importance to the entities. When we put the geometry all in one level, the= =20 circles are as important as the lines and their ID's are sequential. That'= s=20 because the entity needs to have the ability to be persistent. Hmm. Even if it changes from a line to a circle? Layer entities, however, do not need to share a list of ID's with the=20 geometry. They could, but it seems to make more sense that they don't. I guess then that the hierarchy breaks into a few toplevel types. layer object_style text_style mark_style geometry line polyline arc circle mark text image block (?) xref The question about blocks is where their toplevel lives if it's inside of t= he=20 drawing. The block will need an entity like "5.yml" for insert point, etc= =20 (this entity then points to the block (or xref) toplevel.) >do we need separate sub-directories for the header info, or just one? maybe header/layers/, etc Should styles be grouped together? The issue here is that you are likely to get some helpful relationships fro= m=20 having the layer or styles entities in a directory for that type. For=20 instance, you might have a style editor that only opens the files in styles= /. Sorry I can't be more help right now. Looming deadline on Friday. =2D-Eric =2D-=20 "Left to themselves, things tend to go from bad to worse." --Murphy's Corollary