What a full apply changes on the desk

A full atoo apply makes the desk the plan, so nothing from last night’s show is left behind to surprise you. That means it writes more than the values your plan lists: it resets the rest of the desk, unpatches the strips it does not build, and sets a few switches the plan implies. This page lists each of them, so none is a surprise either.

Only a full apply (or the browser’s Apply, which runs it) does any of this. patch, amend and undo touch only what their change owns. Neither does applying a show captured from a desk: it writes only its own patch, routing, names, membership and widths. Every write below is read back and journalled with the value the desk held, so atoo undo reverses it with the rest of the apply.

First, the desk is reset to its initialized state

You do not have to press Initialize on the desk; Atoo does the equivalent itself. It reads the whole desk, compares every setting with an initialized capture of that model (a show file saved from a freshly initialized desk, shipped with Atoo), and writes back the initialized value wherever the desk differs. Then it writes the plan.

A setting the plan writes itself is not reset first: it is written once, as the plan’s value. Some settings are never reset, because they are not part of a show:

Left alone	Why
The desk’s system nodes (`$syscfg`, `$globals`, `$ctl`, and every read-only `$` status value)	IP address, network settings, clock, console name. A write here could cut Atoo’s own connection or the desk’s audio clock.
Expansion cards (`/cards`)	Which card is fitted and how it is clocked.
Meter and RTA display (`/cfg/mtr`, `/cfg/rta`)	What the screen shows, not what the desk does.
Monitor and solo listening (`/cfg/mon`, `/cfg/solo`)	Your headphones and monitor speakers, set for you and the room.

One INFO line counts what was reset, by area. This is a preview of the Demo show against the Demo’s simulated desk:

INFO reset to initialized (WING Compact capture) before the plan's writes: 11 setting(s) would change -- 11 channel; 19,318 already initialized; left alone: 1,648 the plan sets itself, 84 on the skip-list (monitor/solo listening), 26 on the skip-list (meter/RTA display), 18 under a selector the plan sets to another model, 12 on the skip-list (expansion cards), 3 a mono bus's width (the desk holds it at 0), 2 not settable (no address the desk takes); reading the desk took 0.2s (2,121 requests; the plan's own settings: 1,582 from the node dumps, 83 read one by one); writing would take about 0.4s at the rate measured on a WING Compact

The desk is read mostly by node dump, a few requests for hundreds of settings, and a dump prints the desk’s rounded text (144.7 for 144.6788). So a value is judged already initialized from the desk’s rounded readout: one edited by less than the readout’s rounding may be left as it is. That was chosen over reading every number again by itself, which would add thousands of requests to every preview, for a value that is at worst a fraction of a percent off. The settings the reset will change are read exactly, though, because that is the value the journal records and Undo puts back; the line says how many.

The read takes a few seconds. A desk far from initialized can need a few thousand writes, at about 40 ms each.

--no-init skips the reset, for a desk you have initialized by hand. The run notes it at INFO.
A desk whose model Atoo cannot tell (or has no capture for) gets a WARNING, and the plan is written without the reset. Initialize it by hand first. A Compact, a full-size WING and a Rack each have a capture, taken from a real desk of that model.

Strips the plan does not build: unpatched from their sources

A WING ships every channel and aux patched to its own local socket. A full apply unpatches every channel and aux strip the plan does not build: their source goes to OFF, across the whole desk (ch 1-40, aux 1-8). Only the source; the strip’s name, fader, mute and everything else are left as they are.

Without this, a plan that puts the playback on LCL 23-24 would leave ch 24 on its factory patch, LCL 24, listening to the playback. One INFO line says which strips:

INFO ch 24-40, aux 3-7: not in this plan, unpatched from their sources so none listens to a socket the plan uses

Output sockets the plan does not patch: disconnected

A WING ships with its local outputs already patched (on a Compact, outputs 3-8 carry busses and the Mains), and an earlier show leaves its own patches behind. A full apply disconnects every output socket on the desk’s own panel that no bus, Main or Matrix of the plan is patched to, and that the plan does not record as a split: the socket’s group goes to OFF, and nothing else about it is touched. A socket the plan does patch keeps what the plan wrote, and one that already reads OFF is left as it is. One INFO line says which sockets:

INFO output LCL 3-8: no bus or output in this plan is patched there, so each is disconnected rather than left on what the desk came with or an earlier show put there

Without it, the audit would call those patches leftovers, and a stereo bus the plan only feeds into a Main would show as patched on its left side only. This step covers the desk’s own panel. The outputs beyond it (an AES50 stage box, StageCONNECT, a card, USB) are already disconnected by the reset to initialized above, which sets every output group to its initialized state, so nothing needs to be told how many sockets each has. With --no-init, or on a Rack (no capture), --clean-output-sockets reaches them. Undo puts each socket back as it was.

Splits and recording feeds

A plan captured from a desk records every output socket fed straight from a source rather than from a bus, Main or Matrix: a Dante split, the WING-LIVE card one to one, a USB recording feed, an AES50 split (Capturing a desk). Apply writes each one’s source, as it writes a bus’s output patch, so the captured show puts its splits back on any desk. Applied to the desk it came from, they are already there and nothing changes. In the Preview they are on their output group’s tile, beside the buses and Mains patched there.

User signals and user patches: each slot before what reads it

A plan can define the desk’s USR slots: a user signal (USR 1-24) takes a channel, aux, bus, Main or Matrix at its pre or post tap, and a user patch (USR 25-56) takes a physical socket. A channel patched to a slot, or an output fed from one, hears whatever the slot takes, so Apply writes each slot first: what it takes, a user signal’s tap and leg when the plan states them, and its name. A stereo channel on a pair of slots sets both to stereo. Undo puts each slot back as it was. A plan captured from a desk records every slot a channel or an output reads, so applied back to that desk nothing changes; the show’s user.csv says them, one row per slot.

Direct Main routes: every Main is written, OFF included

A WING ships every channel and aux routed straight to Main 1, and bus 9-16 too. So for every strip and bus the plan builds, apply writes all four direct Main routes: on for each Main the plan lists, off for the rest. A drum that should reach the PA only through the Drums bus, or a monitor bus on bus 9-16, is taken off Main 1 rather than reaching it directly as well.

A route that is on is written at unity, post-fader, as is a Matrix’s feed: the conventions have no words for a route’s own level or tap, and the report notes this once.

Talkback: the assign follows the plan

Setup -> Audio’s talkback assign says which strip the desk’s TALK buttons use, and the desk offers only ch 40 or aux 8. When the plan builds a talkback strip on one of those (one named or reading as “Talkback”, “Talk”, “Talk Mic”, “TB” or “T/B”), apply sets the assign to it:

INFO talkback assign (Setup -> Audio) set to AUX8 (was 'OFF'): aux 8 'TALKBACK' is the plan's talkback strip

A plan with no talkback strip there sets the assign to OFF. A talkback strip anywhere else cannot be assigned; verify warns about it (talkback.unassignable) and apply leaves the strip where the plan put it.

Alternate inputs: every strip back on its main input

A channel whose plan states an alternate input (see Writing conventions) gets it written. A channel with no stated alternate keeps whatever the desk has.

Every strip apply builds is also switched back to its main input, alternate or not. A strip left on its alternate after yesterday’s virtual soundcheck hears the recorder, not the stage. The report counts them once:

  - INFO 19 strip(s) set to their main input (not their alternate) -- a show apply means stage inputs live

Apply never writes the desk’s global alternate switch, which moves every strip set to follow it: that is your call at the desk. atoo audit reports it if it is engaged.

Expansion bays: `CRD` and `MOD`

A WING has two slots for extra inputs and outputs: the rear-panel card (CRD: WING-LIVE, Dante, MADI) and an internal AoIP module (MOD: Dante or WSG). Either can be empty, and a desk leaves the factory with nothing in the module slot.

The desk does not tell you. It accepts a patch into an empty bay and reads it back unchanged, so the journal says match and the channel is silent. So apply asks the desk what is fitted before it writes anything, and stops if the plan uses an empty bay:

192.0.2.16 reports NOTHING in the MOD bay (AoIP module: Dante/WSG), and this plan uses it 1 time(s):
  - channel 'Kick' from MOD 1

The console would accept these patches and read them back unchanged -- it does not refuse a patch to an empty bay -- so nothing after this point would tell you the channels are silent. Fit the card or module, or repatch these to a group this desk has.

It exits 2 and nothing is written. --skip-verify does not turn this off. Only the desk answering NONE stops the run: a bay that does not answer, or names a card Atoo does not know, counts as fitted. An empty bay the plan does not use is not a problem.

Offline, atoo verify cannot know what is in a slot, so it warns instead (patch.expansion-bay-not-checked).

Stage boxes: a socket past the end of the box

An AES50 port (A, B, C) lists 48 sockets whatever is plugged into it, so A 17 on an S16 (16 inputs, 8 outputs) is accepted, written and read back by the desk — and the mic on it is heard by nobody. Apply asks the desk what is on each port the plan uses (/$stat/A/dev reads S16) and stops if the plan patches a socket past the end of that box:

This plan patches 1 socket(s) that do not exist on the stage box 192.0.2.16 reports:
  - A 17 (input) -- channel 'Mic 2'; the S16 on AES50-A has 16 inputs, so A 17 does not exist

It exits 2 and nothing is written; --skip-verify does not turn it off. A box Atoo has no size recorded for (only the S16’s is) is reported with a WARNING and its sockets are not checked. A plan that patches no AES50 socket never asks. Patch holds a change to the same rule.

Names go on the source socket

The desk’s Link Customization to Source switch is on by default, and while it is on a strip shows the name, color and icon of the socket it is patched to, not its own. So apply writes the name, color and icon to the source socket, which is where your input list put them. Where one socket feeds more than one strip, those strips are unlinked and carry their own names.

A selector is written before the values it governs

Some settings change what other settings mean. A delay is in meters or feet depending on the strip’s delay mode; a compressor’s threshold means something different under each model. So on each strip apply writes the selectors first, then the values, then the on-switches.

When the run changes a selector, every value it governs is written and read back after it, even if the desk already held the plan’s number: under the old selector that number meant something else. The report says how many at INFO, and the preview counts them the same way.

selector	governs, on the same strip
`in/set/dlymode`	`in/set/dly` only
`dyn/mdl`	every `dyn/*` value
`gate/mdl`	every `gate/*` value
`eq/mdl`, `eq/leq`, `eq/heq`	every `eq/*` value
`flt/mdl`	every `flt/*` value
`gatesc/type`, `dynxo/type`, `dynsc/type`	every value in that node