Nominal Stream Sink#

The block’s ports, parameters and the environment variable it reads.

The block is in the library nominal_lib, listed as Nominal in the Simulink Library Browser. Its label on the canvas shows the mode it sends in: (direct) or (proxy).

Ports#

The block has one input, a vector of doubles whose width equals the number of channel names. The first element is written to the first channel, the second to the second, and so on.

The block has no output unless Add an output port with the discarded-row count is on. The output is then a double scalar: the number of rows dropped so far in the current run.

Parameters#

Text parameters take a MATLAB expression, so enter text in single or double quotes, or the name of a variable that holds the text. No parameter can change while the model runs. The dialog greys out Profile and Profile file in proxy mode, Proxy host and Proxy port in direct mode, and Run title and Ref name when Create run is None.

Channel names (comma-separated): Default 'engine.rpm, engine.egt, engine.psi'. The channel names, in the order of the input’s elements. The number of names sets the input’s width. Spaces around each name are ignored, and so are empty entries such as a trailing comma.
Detect from connected signal: A button. Fills in the channel names from whatever is wired into the block. Choose and name channels lists where the names come from.
Dataset RID: Default ''. The RID of the dataset to write to. The dataset has to exist already. Required.
Tags (key and value pairs): Default [], no tags. Tags on every channel the block writes, which keep the runs of one model apart in a dataset they share. Enter a two-column string array with one row per tag, such as ["model" "A"; "execution-id" "7"], a cell array of the same shape, a dictionary or a containers.Map. A value can be a number. Keys can’t be empty or repeat. See Simulation before hardware for how tags and runs fit together.
Execution ID: Default 'auto'. Sent as the tag execution-id on every channel the block writes, so each run can be picked out of the dataset. 'auto' makes a new ID when each run starts, such as 20261019-141048-3f2a9c: the run’s start time in UTC, then six hexadecimal digits that differ between runs starting in the same second. Text or a number sets the ID yourself, such as a variable a script sets for each run. '' sends no execution ID. Tags (key and value pairs) can’t have an execution-id key unless this field is empty. See Run a campaign of simulations.
Mode: Default Direct to Nominal. Direct to Nominal sends over HTTPS from the computer running the model. Via proxy sends over UDP to nmproxy, which writes to Nominal. See Stream through a proxy.
Profile: Default 'default'. Direct mode only. The profile in Nominal’s config file to read the API key and the instance’s REST API base URL from, as its token and base_url. It is the same file Nominal’s other clients read. API keys setup shows what the file holds.
Profile file: Default '', the standard location: ~/.config/nominal/config.yml, or %USERPROFILE%\.config\nominal\config.yml on Windows. Direct mode only. The path of the config file to read Profile from, when it is somewhere else.
Create run: Default None. One per execution creates a run in Nominal when each run of the model starts, so it can be watched live, and sets the run’s end once the last row is sent. The run shows the dataset’s points with this execution’s tags, Execution ID included. One per model shares one run between every execution with the same title: the first execution creates it, later ones extend its end, and it shows every point in the dataset, so every execution’s points are in it whatever their tags. Finding the run means searching every run whose title contains this one, so the more runs have similar titles, the longer an execution takes to start. In direct mode, executions that start at the same moment, such as parsim workers, can each create the run. When a run can’t be created or ended, the block warns and the rows still go to the dataset. In proxy mode, nmproxy creates and ends the run with its own profile, and a run it can’t create or end shows in its log instead.
Run title: Default '', the model’s name. One per execution adds the execution ID, as in Model A, execution 7, or with no execution ID the UTC time the execution started, so each execution’s run has a title of its own: 'Bench test' gives Bench test, execution 7. One per model uses the title as it is. Used only with Create run.
Ref name: Default 'default'. The name the dataset goes under in the run. Workbooks find a channel by its ref name and channel name together, so use the ref name your other runs use, such as the hardware team’s. Used only with Create run.
Proxy host: Default '127.0.0.1'. Proxy mode only. The name or IP address of the computer running nmproxy.
Proxy port: Default 15400. Proxy mode only. The UDP port nmproxy listens on, from 1 to 65535.
Sample time (s) — how often to write, not the solver step: Default 0.001. Seconds between rows, greater than 0. Simulink’s -1, which inherits a sample time, isn’t accepted. See Timing and dropped rows.
Buffer capacity (rows): Default 4096. How many rows the block’s buffer holds while they wait to be sent, as a whole number greater than 0. The buffer is reserved in memory when the run starts, and rows arriving while it is full are dropped. When the run ends, the block spends up to 15 seconds sending what is left in the buffer, and counts any rows still unsent as dropped.
Maximum delay before sending (s): Default 1. The longest a row waits in the buffer before the block sends it, in seconds greater than 0. The block sends sooner as rows arrive faster: in direct mode once 4096 rows are waiting, or half the buffer if that is less, and in proxy mode once a packet’s worth is, as The frame protocol sizes it. See Timing and dropped rows.
Timestamps: Default Simulation time from start. Simulation time from start stamps each row with the time the run started plus its simulation time. Directly record simulation time (t0=0) stamps each row with its simulation time alone, counted from the Unix epoch, so every run starts at the same instant. Choosing it while Tags (key and value pairs) and Execution ID are both empty asks you to confirm, and Cancel sets Simulation time from start. Host wall clock stamps each row with the computer’s clock when the block writes it. See Timing and dropped rows.
Add an output port with the discarded-row count: Default off. Adds the output described under Ports.

Environment variable#

The block reads this when a run starts, in direct mode only.

NOMINAL_CA_BUNDLE: Optional. The path to a PEM file of CA certificates to verify Nominal’s HTTPS certificate with. Set it when your Nominal instance or your network uses certificates that the operating system doesn’t trust. Unset or empty uses the operating system’s certificates.

Limits in proxy mode#

Each run’s channel list travels to nmproxy in a single packet, which holds at most 1400 bytes. The dataset RID, the channel names, the tags, and with Create run on the run’s title and Ref name, have to fit:

45 + (RID length) + (run title length) + (ref name length) + the sum over channels of (2 + name length) + the sum over tags of (4 + key length + value length) ≤ 1400

Lengths are in bytes, and the run title and ref name count as 0 when Create run is None. Execution ID counts as a tag with the key execution-id, and an automatic one is 22 characters long. With a 70-character RID, 10-character names, an automatic execution ID as the only tag and no run, that is 103 channels. The RID, each name, each tag key and value, the run title and the ref name can each be at most 256 bytes. A block can have at most 64 tags, and at most 169 channels however short the names are. The frame protocol has the reasoning behind these limits.

A block that breaks one of the limits stops when the run starts, with in proxy mode, the dataset RID, channel names, tags and run title have to fit in one 1400-byte packet to the relay, and these don't.