← All demos

Language & UI

Layers: a rig schematic

describe a symbol once, place it forty times.

catalog/qu_layers_schematic.qu

A measurement rig drawn as a schematic: sensors along a beam, an amplifier per channel, and the wiring between them. Every symbol is a LAYER -- captured once from a function that draws it at the origin, then stamped wherever it belongs, at whatever size and angle.

WHY THAT MATTERS. Without layers this script is the same eight drawing calls copied fourteen times with different numbers in them, and changing the sensor symbol means finding all fourteen. With them the symbol has one definition and the placement is a loop, which is the difference between a figure you can edit and one you redraw.

A LAYER IS A VALUE, not a mode with a begin and an end. It can be passed to a function, returned from one, kept in a list, and duplicated by ordinary assignment -- which is what lets `symbols` below be a list of them, indexed by which channel is being drawn.

What it exercises

• capture(fn, [args...]) — run a function, keep what it drew
• stamp(layer, x, y, scale=, rotate=)
• circle / rect / polygon / arc — the symbols themselves
• hsv / palette — colour picked by role, not by taste
• axis off — the panel as a coordinate system

How it works

function sensor()
    circle(0, 0, 0.30, fill = LIVE, alpha = 0.85)
    circle(0, 0, 0.46, color = LIVE, width = 1.4)
    # The lead, pointing down toward the beam.
    plot([0, 0], [-0.46, -1.0], color = LIVE, width = 1.4)
end function

function amplifier()
    # The triangle every schematic uses, pointing right.
    polygon((-0.45, 0.55, -0.45), (-0.5, 0.0, 0.5), fill = "#ffffff", alpha = 1.0)
    polygon((-0.45, 0.55, -0.45), (-0.5, 0.0, 0.5), color = INK, width = 1.3)
    plot([-0.85, -0.45], [0, 0], color = INK, width = 1.2)
    plot([0.55, 0.95], [0, 0], color = INK, width = 1.2)
end function

function adc()
    rect(-0.5, -0.4, 0.5, 0.4, color = INK)
    plot([-0.9, -0.5], [0, 0], color = INK, width = 1.2)
    plot([0.5, 0.9], [0, 0], color = INK, width = 1.2)
    # The staircase that says "sampled".
    plot([-0.30, -0.15, -0.15, 0.00, 0.00, 0.15, 0.15, 0.30],
         [-0.18, -0.18, -0.02, -0.02, 0.10, 0.10, 0.22, 0.22],
         color = WARM, width = 1.2)
end function

function fault()
    # A warning marker, for the channel that is misbehaving. Drawn at unit
    # size like everything else, so `scale=` alone sizes it.
    polygon((0, 0.5, -0.5), (0.55, -0.35, -0.35), fill = "#f6d6c0", alpha = 1.0)
    polygon((0, 0.5, -0.5), (0.55, -0.35, -0.35), color = WARM, width = 1.3)
    plot([0, 0], [-0.08, 0.28], color = WARM, width = 1.6)
    circle(0, -0.20, 0.045, fill = WARM, alpha = 1.0)
end function

SENSOR = capture("sensor")
AMP    = capture("amplifier")
ADC    = capture("adc")
FAULT  = capture("fault")
print("captured: {SENSOR}, {AMP}, {ADC}, {FAULT}")

N = 7
BEAM_Y = 1.6

figure_size(1100, 560)
xlim(-0.5, 15.5)
ylim(0, 9.2)

# The beam the sensors sit on.
rect(0.2, BEAM_Y - 0.30, 14.8, BEAM_Y + 0.30, color = "#c8ccd4")
text(0.35, BEAM_Y - 0.75, "test beam", size = 11, color = "#59616f")

# One channel per sensor: sensor, wire, amplifier, wire, converter.
for k in 0 to N - 1
    x = 1.4 + k * 2.0
    stamp(SENSOR, x, BEAM_Y + 1.6)

    # The wire from the sensor up to its amplifier.
    plot([x, x], [BEAM_Y + 2.6, 4.7], color = "#9aa3b0", width = 1.0)

    stamp(AMP, x, 5.3, scale = 0.78)
    plot([x, x], [5.9, 6.6], color = "#9aa3b0", width = 1.0)
    stamp(ADC, x, 7.2, scale = 0.78)

    text(x, 8.1, "ch {k}", size = 11, color = "#59616f")
end for

# Channel 4 has a problem. One extra stamp says so -- no new symbol, no
# copied code, just the same layer placed somewhere with a reason.
stamp(FAULT, 1.4 + 4 * 2.0 + 0.78, 5.9, scale = 0.62)
text(1.4 + 4 * 2.0 + 1.5, 5.9, "clipping", size = 11, color = WARM)

# The bus every converter feeds.
plot([1.4 - 0.7, 1.4 + (N - 1) * 2.0 + 0.7], [8.7, 8.7], color = INK, width = 1.4)
for k in 0 to N - 1
    x = 1.4 + k * 2.0
    plot([x, x], [7.6, 8.7], color = "#9aa3b0", width = 1.0)
end for
text(1.4 + (N - 1) * 2.0 + 0.9, 8.7, "to host", size = 11, color = INK)

title("Seven channels, four symbols, one definition each")
axis off
savefig("layers_schematic.svg")

print("{N} channels drawn from 4 captured symbols")
print("stamps placed: {N * 3 + 1}")

Results

captured: layer(2 shapes, 1 series), layer(2 shapes, 2 series), layer(1 shape, 3 series), layer(3 shapes, 1 series)
  note: this figure uses screen styling and may not hold up in print -- call theme("publication") before saving for a paper (said once per session)
  note: fonts render using whatever is installed on the viewer's machine (none are embedded) -- pass embed_fonts=true (fontfamily preset only) or export .tikz for guaranteed appearance
7 channels drawn from 4 captured symbols
stamps placed: 22

The whole program

Runnable as it stands: qu run catalog/qu_layers_schematic.qu

# qu_layers_schematic.qu — describe a symbol once, place it forty times.

#

# A measurement rig drawn as a schematic: sensors along a beam, an

# amplifier per channel, and the wiring between them. Every symbol is a

# LAYER -- captured once from a function that draws it at the origin, then

# stamped wherever it belongs, at whatever size and angle.

#

# WHY THAT MATTERS. Without layers this script is the same eight drawing

# calls copied fourteen times with different numbers in them, and changing

# the sensor symbol means finding all fourteen. With them the symbol has one

# definition and the placement is a loop, which is the difference between a

# figure you can edit and one you redraw.

#

# A LAYER IS A VALUE, not a mode with a begin and an end. It can be passed

# to a function, returned from one, kept in a list, and duplicated by

# ordinary assignment -- which is what lets `symbols` below be a list of

# them, indexed by which channel is being drawn.

#

# Exercises:

#   capture(fn, [args...])         — run a function, keep what it drew

#   stamp(layer, x, y, scale=, rotate=)

#   circle / rect / polygon / arc  — the symbols themselves

#   hsv / palette                  — colour picked by role, not by taste

#   axis off                       — the panel as a coordinate system


backend auto

INK  = "#26303f"
LIVE = "#0072bd"
WARM = "#b5651d"

#%% The symbols

#

# Each is drawn at the origin, pointing right, at unit size. Everything

# about WHERE it goes belongs to the stamp, not to the symbol -- that

# separation is the whole point.


function sensor()
    circle(0, 0, 0.30, fill = LIVE, alpha = 0.85)
    circle(0, 0, 0.46, color = LIVE, width = 1.4)
    # The lead, pointing down toward the beam.

    plot([0, 0], [-0.46, -1.0], color = LIVE, width = 1.4)
end function

function amplifier()
    # The triangle every schematic uses, pointing right.

    polygon((-0.45, 0.55, -0.45), (-0.5, 0.0, 0.5), fill = "#ffffff", alpha = 1.0)
    polygon((-0.45, 0.55, -0.45), (-0.5, 0.0, 0.5), color = INK, width = 1.3)
    plot([-0.85, -0.45], [0, 0], color = INK, width = 1.2)
    plot([0.55, 0.95], [0, 0], color = INK, width = 1.2)
end function

function adc()
    rect(-0.5, -0.4, 0.5, 0.4, color = INK)
    plot([-0.9, -0.5], [0, 0], color = INK, width = 1.2)
    plot([0.5, 0.9], [0, 0], color = INK, width = 1.2)
    # The staircase that says "sampled".

    plot([-0.30, -0.15, -0.15, 0.00, 0.00, 0.15, 0.15, 0.30],
         [-0.18, -0.18, -0.02, -0.02, 0.10, 0.10, 0.22, 0.22],
         color = WARM, width = 1.2)
end function

function fault()
    # A warning marker, for the channel that is misbehaving. Drawn at unit

    # size like everything else, so `scale=` alone sizes it.

    polygon((0, 0.5, -0.5), (0.55, -0.35, -0.35), fill = "#f6d6c0", alpha = 1.0)
    polygon((0, 0.5, -0.5), (0.55, -0.35, -0.35), color = WARM, width = 1.3)
    plot([0, 0], [-0.08, 0.28], color = WARM, width = 1.6)
    circle(0, -0.20, 0.045, fill = WARM, alpha = 1.0)
end function

#%% Capture each one, once

SENSOR = capture("sensor")
AMP    = capture("amplifier")
ADC    = capture("adc")
FAULT  = capture("fault")
print("captured: {SENSOR}, {AMP}, {ADC}, {FAULT}")

#%% The rig

N = 7
BEAM_Y = 1.6

figure_size(1100, 560)
xlim(-0.5, 15.5)
ylim(0, 9.2)

# The beam the sensors sit on.

rect(0.2, BEAM_Y - 0.30, 14.8, BEAM_Y + 0.30, color = "#c8ccd4")
text(0.35, BEAM_Y - 0.75, "test beam", size = 11, color = "#59616f")

# One channel per sensor: sensor, wire, amplifier, wire, converter.

for k in 0 to N - 1
    x = 1.4 + k * 2.0
    stamp(SENSOR, x, BEAM_Y + 1.6)

    # The wire from the sensor up to its amplifier.

    plot([x, x], [BEAM_Y + 2.6, 4.7], color = "#9aa3b0", width = 1.0)

    stamp(AMP, x, 5.3, scale = 0.78)
    plot([x, x], [5.9, 6.6], color = "#9aa3b0", width = 1.0)
    stamp(ADC, x, 7.2, scale = 0.78)

    text(x, 8.1, "ch {k}", size = 11, color = "#59616f")
end for

# Channel 4 has a problem. One extra stamp says so -- no new symbol, no

# copied code, just the same layer placed somewhere with a reason.

stamp(FAULT, 1.4 + 4 * 2.0 + 0.78, 5.9, scale = 0.62)
text(1.4 + 4 * 2.0 + 1.5, 5.9, "clipping", size = 11, color = WARM)

# The bus every converter feeds.

plot([1.4 - 0.7, 1.4 + (N - 1) * 2.0 + 0.7], [8.7, 8.7], color = INK, width = 1.4)
for k in 0 to N - 1
    x = 1.4 + k * 2.0
    plot([x, x], [7.6, 8.7], color = "#9aa3b0", width = 1.0)
end for
text(1.4 + (N - 1) * 2.0 + 0.9, 8.7, "to host", size = 11, color = INK)

title("Seven channels, four symbols, one definition each")
axis off
savefig("layers_schematic.svg")

#%% What it cost

#

# Four symbol definitions and one loop. Placed by hand this figure is

# roughly 7 x 8 drawing calls with different numbers in each -- and moving

# the beam would mean editing every one of them.

print("{N} channels drawn from 4 captured symbols")
print("stamps placed: {N * 3 + 1}")