@Section
@Tag { dia_posi }
@Title { Positioning }
@Begin
@PP
Once the nodes of the diagram are in place, @@Diag can be trusted to look
diagrams. @RawIndex { diagrams }
diagrams.positioning @SubIndex { positioning nodes }
positioning.diagrams @Index { positioning nodes in diagrams }
after the rest: links to standard outlines will terminate neatly on their
boundaries, labels will not overstrike links no matter what direction they
are heading, and so on. The great weakness of @@Diag is in positioning
the nodes. This is partly because `what pleases the eye' is the
positioning rule in many diagrams, and an interactive system is really
needed in such cases; and partly because, even when the rule is more formal
(for example, when the nodes are to be laid out in a grid), @@Diag does not
have symbols to produce it anyway.
@PP
Previous examples have used @Code "@DP" for getting nodes one under
another, and white space between nodes for getting them side by side, but
this is very primitive. This section suggests three better ways: using
{@Code "@Tbl"}, using {@Code "@Graph"}, and using coordinates; and the
following section adds a fourth, using @@Diag's tree-drawing symbols. It's
a bit of a jumble.
@PP
The {@Code "@Tbl"} symbol (Chapter {@NumberOf tables}) is a good choice when
the nodes have any kind of grid-like arrangement:
@ID @OneRow {
@Code @Verbatim {
@Diag {
@Tbl
aformat { @Cell A | @Cell B | @Cell C }
marginhorizontal { 0.5c }
marginvertical { 0.25c }
{
@Rowa
B { A:: @Square }
@Rowa
A { B:: @Square }
C { C:: @Square }
@Rowa
B { D:: @Square }
}
//
@Arrow from { A } to { B }
@Arrow from { A } to { C }
@Arrow from { B } to { D }
@Arrow from { C } to { D }
@Arrow from { A } to { D }
}
}
||9ct
@Diag {
@Tbl
aformat { @Cell A | @Cell B | @Cell C }
marginhorizontal { 0.5c }
marginvertical { 0.25c }
{
@Rowa
B { A:: @Square }
@Rowa
A { B:: @Square }
C { C:: @Square }
@Rowa
B { D:: @Square }
}
//
@Arrow from { A } to { B }
@Arrow from { A } to { C }
@Arrow from { B } to { D }
@Arrow from { C } to { D }
@Arrow from { A } to { D }
}
}
The table occupies the nodes part. Tags may have the same name
as columns; the two can never conflict.
@PP
Similarly, the @Code "@Graph" symbol from Chapter {@NumberOf graphs}
has an @Code "objects" option which can place arbitrary objects,
including labelled nodes, anywhere on a graph:
@ID @OneRow {
@Code @Verbatim {
@Diag {
@Graph
xmin { 0 }
xmax { 100 }
ymin { 0 }
ymax { 100 }
objects {
@CTR at { 20 30 } { A:: @Square }
@CTR at { 60 70 } { B:: @Square }
}
{}
//
@Link from { A } to { B }
}
}
||8.5ct
@Diag {
@Graph
xmin { 0 }
xmax { 100 }
ymin { 0 }
ymax { 100 }
objects {
@CTR at { 20 30 } { A:: @Square }
@CTR at { 60 70 } { B:: @Square }
}
{}
//
@Link from { A } to { B }
}
}
Once again the @Code "@Graph" symbol occupies the nodes part. You can
get rid of the axes by setting the @Code "style" option of @Code "@Graph"
to {@Code none}, and then it won't look like a graph at all.
@PP
@@Diag has a system of node positioning based on coordinates which is
somewhat similar to the @Code "@Graph" one. It is often the easiest way
to scatter nodes about a diagram at random. The first step is to create
a nodes part that is just an empty space of whatever size you want the
final diagram to be:
@ID @OneRow @Code @Verbatim {
@Diag {
4c @High 6c @Wide
//
...
}
}
As shown, this is done with the @Code "@Wide" and @Code "@High" symbols
from basic Lout; the above diagram will be four centimetres high by
six centimetres wide.
@PP
@@Diag has a @Code "," symbol that allows you to specify a point by
diagrams. @RawIndex { diagrams }
diagrams.coordinates @SubIndex { coordinates }
coordinates.diagrams @Index { coordinates in diagrams }
its coordinates in the diagram's base. For example,
@Code "0,0" denotes the bottom left-hand corner of the base,
@Code "1,0" denotes the bottom right-hand corner, and
@Code "0.5,0.5" denotes the centre of the base. Coordinates should
usually be between 0 and 1, since otherwise they denote points
outside the base (which is allowed but seldom useful).
@PP
Every node symbol has a @Code "translate" option which allows you
diagrams. @RawIndex { diagrams }
diagrams.translate @SubIndex { @Code "translate" option }
translate.diagrams @Index { @Code "translate" option (diagrams) }
to move the node about on the diagram's base (or off it if you use
coordinates less than 0 or greater than 1). If you use this option,
the node effectively has zero size and overstrikes anything else
in the area you put it (like labels do). It is best to put these
nodes in the links part:
@ID @OneRow {
@Code @Verbatim {
@Diag {
@Box margin { 0c } 4c @Wide 5c @High
//
A:: @Square
translate { CTR to 0.5, 0.67 }
{ @I A }
B:: @Circle
translate { CTR to 0.8, 0.25 }
{ @I B }
}
}
||9ct
@Diag {
@Box margin { 0c } 4c @Wide 5c @High
//
A:: @Square
translate { CTR to 0.5, 0.67 }
{ @I A }
B:: @Circle
translate { CTR to 0.8, 0.25 }
{ @I B }
}
}
A box with margin zero has been drawn around the empty space to
show its extent. The value of @Code "translate" should always
be {@I point} @Code to {@I point}; the first point lies within
the node, the second lies within the nodes part, and the translation
makes these two points coincide.
@PP
You are free to have nodes in the nodes part as well, or any object
at all. Here is an example which shows what a little ingenuity
can accomplish:
@ID @OneRow {
@Code @Verbatim {
@Diag {
@Polygon
sides { 5 }
outlinestyle { noline }
hsize { 4c }
vsize { 4c }
//
A:: @Circle translate { N to P1 } {}
B:: @Circle translate { N to P2 } {}
C:: @Circle translate { N to P3 } {}
D:: @Circle translate { N to P4 } {}
E:: @Circle translate { N to P5 } {}
@Link arrow { both } from { A } to { B }
@Link arrow { both } from { B } to { C }
@Link arrow { both } from { C } to { D }
@Link arrow { both } from { D } to { E }
@Link arrow { both } from { E } to { A }
}
}
||9ct
@Diag {
@Polygon
sides { 5 }
outlinestyle { noline }
hsize { 4c }
vsize { 4c }
//
A:: @Circle translate { N to P1 } {}
B:: @Circle translate { N to P2 } {}
C:: @Circle translate { N to P3 } {}
D:: @Circle translate { N to P4 } {}
E:: @Circle translate { N to P5 } {}
@Link arrow { both } from { A } to { B }
@Link arrow { both } from { B } to { C }
@Link arrow { both } from { C } to { D }
@Link arrow { both } from { D } to { E }
@Link arrow { both } from { E } to { A }
}
}
This uses the tags of a phantom polygon to position the
real nodes. It would be a rare interactive system that could
position nodes with this precision; @@Diag shines whenever there
is a formal positioning rule to follow.
@End @Section
