test(e2e): avoid timing issues in rpc_gui_obj test

Automatically generated by python-semantic-release

CHANGELOG.md

@@ -1,6 +1,17 @@
 # CHANGELOG
 
 
+## v3.2.0 (2026-03-11)
+
+### Features
+
+- **curve, waveform**: Add dap_parameters for lmfit customization in DAP requests
+  ([`14d51b8`](https://github.com/bec-project/bec_widgets/commit/14d51b80169f5a060dd24287f3a6db9a4b41275a))
+
+- **waveform**: Composite DAP with multiple models
+  ([`b4f6f5a`](https://github.com/bec-project/bec_widgets/commit/b4f6f5aa8bcd0f6091610e8f839ea265c87575e0))
+
+
 ## v3.1.4 (2026-03-11)
 
 ### Bug Fixes

bec_widgets/cli/client.py

@@ -6249,7 +6249,8 @@ class Waveform(RPCBase):
         signal_y: "str | None" = None,
         color: "str | None" = None,
         label: "str | None" = None,
-        dap: "str | None" = None,
+        dap: "str | list[str] | None" = None,
+        dap_parameters: "dict | list | lmfit.Parameters | None | object" = None,
         scan_id: "str | None" = None,
         scan_number: "int | None" = None,
         **kwargs,
@@ -6271,9 +6272,14 @@ class Waveform(RPCBase):
             signal_y(str): The name of the entry for the y-axis.
             color(str): The color of the curve.
             label(str): The label of the curve.
-            dap(str): The dap model to use for the curve. When provided, a DAP curve is
+            dap(str | list[str]): The dap model to use for the curve. When provided, a DAP curve is
                 attached automatically for device, history, or custom data sources. Use
-                the same string as the LMFit model name.
+                the same string as the LMFit model name, or a list of model names to build a composite.
+            dap_parameters(dict | list | lmfit.Parameters | None): Optional lmfit parameter overrides sent to
+                the DAP server. For a single model: values can be numeric (interpreted as fixed parameters)
+                or dicts like `{"value": 1.0, "vary": False}`. For composite models (dap is list), use either
+                a list aligned to the model list (each item is a param dict), or a dict of
+                `{ "ModelName": { "param": {...} } }` when model names are unique.
             scan_id(str):  Optional scan ID. When provided, the curve is treated as a **history** curve and
                 the y‑data (and optional x‑data) are fetched from that historical scan. Such curves are
                 never cleared by live‑scan resets.
@@ -6287,9 +6293,10 @@ class Waveform(RPCBase):
     def add_dap_curve(
         self,
         device_label: "str",
-        dap_name: "str",
+        dap_name: "str | list[str]",
         color: "str | None" = None,
         dap_oversample: "int" = 1,
+        dap_parameters: "dict | list | lmfit.Parameters | None" = None,
         **kwargs,
     ) -> "Curve":
         """
@@ -6299,9 +6306,11 @@ class Waveform(RPCBase):
 
         Args:
             device_label(str): The label of the source curve to add DAP to.
-            dap_name(str): The name of the DAP model to use.
+            dap_name(str | list[str]): The name of the DAP model to use, or a list of model
+                names to build a composite model.
             color(str): The color of the curve.
             dap_oversample(int): The oversampling factor for the DAP curve.
+            dap_parameters(dict | list | lmfit.Parameters | None): Optional lmfit parameter overrides sent to the DAP server.
             **kwargs
 
         Returns:

bec_widgets/widgets/plots/waveform/curve.py

@@ -22,8 +22,9 @@ class DeviceSignal(BaseModel):
 
     device: str
     signal: str
-    dap: str | None = None
+    dap: str | list[str] | None = None
     dap_oversample: int = 1
+    dap_parameters: dict | list | None = None
 
     model_config: dict = {"validate_assignment": True}
 

bec_widgets/widgets/plots/waveform/waveform.py

@@ -1,13 +1,13 @@
 from __future__ import annotations
 
 import json
-from typing import Literal
+from typing import TYPE_CHECKING, Literal
 
-import lmfit
 import numpy as np
 import pyqtgraph as pg
 from bec_lib import bec_logger, messages
 from bec_lib.endpoints import MessageEndpoints
+from bec_lib.lmfit_serializer import serialize_lmfit_params, serialize_param_object
 from bec_lib.scan_data_container import ScanDataContainer
 from pydantic import Field, ValidationError, field_validator
 from qtpy.QtCore import Qt, QTimer, Signal
@@ -41,6 +41,18 @@ from bec_widgets.widgets.services.scan_history_browser.scan_history_browser impo
 )
 
 logger = bec_logger.logger
+_DAP_PARAM = object()
+
+if TYPE_CHECKING:  # pragma: no cover
+    import lmfit  # type: ignore
+else:
+    try:
+        import lmfit  # type: ignore
+    except Exception as e:  # pragma: no cover
+        logger.warning(
+            f"lmfit could not be imported: {e}. Custom DAP functionality will be unavailable."
+        )
+        lmfit = None
 
 
 # noinspection PyDataclass
@@ -696,7 +708,8 @@ class Waveform(PlotBase):
         signal_y: str | None = None,
         color: str | None = None,
         label: str | None = None,
-        dap: str | None = None,
+        dap: str | list[str] | None = None,
+        dap_parameters: dict | list | lmfit.Parameters | None | object = None,
         scan_id: str | None = None,
         scan_number: int | None = None,
         **kwargs,
@@ -718,9 +731,14 @@ class Waveform(PlotBase):
             signal_y(str): The name of the entry for the y-axis.
             color(str): The color of the curve.
             label(str): The label of the curve.
-            dap(str): The dap model to use for the curve. When provided, a DAP curve is
+            dap(str | list[str]): The dap model to use for the curve. When provided, a DAP curve is
                 attached automatically for device, history, or custom data sources. Use
-                the same string as the LMFit model name.
+                the same string as the LMFit model name, or a list of model names to build a composite.
+            dap_parameters(dict | list | lmfit.Parameters | None): Optional lmfit parameter overrides sent to
+                the DAP server. For a single model: values can be numeric (interpreted as fixed parameters)
+                or dicts like `{"value": 1.0, "vary": False}`. For composite models (dap is list), use either
+                a list aligned to the model list (each item is a param dict), or a dict of
+                `{ "ModelName": { "param": {...} } }` when model names are unique.
             scan_id(str):  Optional scan ID. When provided, the curve is treated as a **history** curve and
                 the y‑data (and optional x‑data) are fetched from that historical scan. Such curves are
                 never cleared by live‑scan resets.
@@ -733,6 +751,8 @@ class Waveform(PlotBase):
         source = "custom"
         x_data = None
         y_data = None
+        if dap_parameters is _DAP_PARAM:
+            dap_parameters = kwargs.pop("dap_parameters", None) or kwargs.pop("parameters", None)
 
         # 1. Custom curve logic
         if x is not None and y is not None:
@@ -810,7 +830,9 @@ class Waveform(PlotBase):
         curve = self._add_curve(config=config, x_data=x_data, y_data=y_data)
 
         if dap is not None and curve.config.source in ("device", "history", "custom"):
-            self.add_dap_curve(device_label=curve.name(), dap_name=dap, **kwargs)
+            self.add_dap_curve(
+                device_label=curve.name(), dap_name=dap, dap_parameters=dap_parameters, **kwargs
+            )
 
         return curve
 
@@ -820,9 +842,10 @@ class Waveform(PlotBase):
     def add_dap_curve(
         self,
         device_label: str,
-        dap_name: str,
+        dap_name: str | list[str],
         color: str | None = None,
         dap_oversample: int = 1,
+        dap_parameters: dict | list | lmfit.Parameters | None = None,
         **kwargs,
     ) -> Curve:
         """
@@ -832,9 +855,11 @@ class Waveform(PlotBase):
 
         Args:
             device_label(str): The label of the source curve to add DAP to.
-            dap_name(str): The name of the DAP model to use.
+            dap_name(str | list[str]): The name of the DAP model to use, or a list of model
+                names to build a composite model.
             color(str): The color of the curve.
             dap_oversample(int): The oversampling factor for the DAP curve.
+            dap_parameters(dict | list | lmfit.Parameters | None): Optional lmfit parameter overrides sent to the DAP server.
             **kwargs
 
         Returns:
@@ -859,7 +884,7 @@ class Waveform(PlotBase):
             dev_entry = "custom"
 
         # 2) Build a label for the new DAP curve
-        dap_label = f"{device_label}-{dap_name}"
+        dap_label = f"{device_label}-{self._format_dap_label(dap_name)}"
 
         # 3) Possibly raise if the DAP curve already exists
         if self._check_curve_id(dap_label):
@@ -882,7 +907,11 @@ class Waveform(PlotBase):
 
         # Attach device signal with DAP
         config.signal = DeviceSignal(
-            device=dev_name, signal=dev_entry, dap=dap_name, dap_oversample=dap_oversample
+            device=dev_name,
+            signal=dev_entry,
+            dap=dap_name,
+            dap_oversample=dap_oversample,
+            dap_parameters=self._normalize_dap_parameters(dap_parameters, dap_name=dap_name),
         )
 
         # 4) Create the DAP curve config using `_add_curve(...)`
@@ -1754,7 +1783,9 @@ class Waveform(PlotBase):
 
             x_data, y_data = parent_curve.get_data()
             model_name = dap_curve.config.signal.dap
-            model = getattr(self.dap, model_name)
+            model = None
+            if not isinstance(model_name, (list, tuple)):
+                model = getattr(self.dap, model_name)
             try:
                 x_min, x_max = self.roi_region
                 x_data, y_data = self._crop_data(x_data, y_data, x_min, x_max)
@@ -1762,20 +1793,132 @@ class Waveform(PlotBase):
                 x_min = None
                 x_max = None
 
+            dap_parameters = getattr(dap_curve.config.signal, "dap_parameters", None)
+            dap_kwargs = {
+                "data_x": x_data,
+                "data_y": y_data,
+                "oversample": dap_curve.dap_oversample,
+            }
+            if dap_parameters:
+                dap_kwargs["parameters"] = dap_parameters
+
+            if model is not None:
+                class_args = model._plugin_info["class_args"]
+                class_kwargs = model._plugin_info["class_kwargs"]
+            else:
+                class_args = []
+                class_kwargs = {"model": model_name}
+
             msg = messages.DAPRequestMessage(
                 dap_cls="LmfitService1D",
                 dap_type="on_demand",
                 config={
                     "args": [],
-                    "kwargs": {"data_x": x_data, "data_y": y_data},
-                    "class_args": model._plugin_info["class_args"],
-                    "class_kwargs": model._plugin_info["class_kwargs"],
+                    "kwargs": dap_kwargs,
+                    "class_args": class_args,
+                    "class_kwargs": class_kwargs,
                     "curve_label": dap_curve.name(),
                 },
                 metadata={"RID": f"{self.scan_id}-{self.gui_id}"},
             )
             self.client.connector.set_and_publish(MessageEndpoints.dap_request(), msg)
 
+    @staticmethod
+    def _normalize_dap_parameters(
+        parameters: dict | list | lmfit.Parameters | None, dap_name: str | list[str] | None = None
+    ) -> dict | list | None:
+        """
+        Normalize user-provided lmfit parameters into a JSON-serializable dict suitable for the DAP server.
+
+        Supports:
+        - `lmfit.Parameters` (single-model only)
+        - `dict[name -> number]` (treated as fixed parameter with `vary=False`)
+        - `dict[name -> dict]` (lmfit.Parameter fields; defaults to `vary=False` if unspecified)
+        - `dict[name -> lmfit.Parameter]`
+        - composite: `list[dict[param_name -> spec]]` aligned to model list
+        - composite: `dict[model_name -> dict[param_name -> spec]]` (unique model names only)
+        """
+        if parameters is None:
+            return None
+        if isinstance(dap_name, (list, tuple)):
+            if lmfit is not None and isinstance(parameters, lmfit.Parameters):
+                raise TypeError("dap_parameters must be a dict when using composite dap models.")
+            if isinstance(parameters, (list, tuple)):
+                normalized_list: list[dict | None] = []
+                for idx, item in enumerate(parameters):
+                    if item is None:
+                        normalized_list.append(None)
+                        continue
+                    if not isinstance(item, dict):
+                        raise TypeError(
+                            f"dap_parameters list item {idx} must be a dict of parameter overrides."
+                        )
+                    normalized_list.append(Waveform._normalize_param_overrides(item))
+                return normalized_list or None
+            if not isinstance(parameters, dict):
+                raise TypeError(
+                    "dap_parameters must be a dict of model->params when using composite dap models."
+                )
+            model_names = set(dap_name)
+            invalid_models = set(parameters.keys()) - model_names
+            if invalid_models:
+                raise TypeError(
+                    f"Invalid dap_parameters keys for composite model: {sorted(invalid_models)}"
+                )
+            normalized_composite: dict[str, dict] = {}
+            for model_name in dap_name:
+                model_params = parameters.get(model_name)
+                if model_params is None:
+                    continue
+                if not isinstance(model_params, dict):
+                    raise TypeError(
+                        f"dap_parameters for '{model_name}' must be a dict of parameter overrides."
+                    )
+                normalized = Waveform._normalize_param_overrides(model_params)
+                if normalized:
+                    normalized_composite[model_name] = normalized
+            return normalized_composite or None
+
+        if lmfit is not None and isinstance(parameters, lmfit.Parameters):
+            return serialize_lmfit_params(parameters)
+        if not isinstance(parameters, dict):
+            if lmfit is None:
+                raise TypeError(
+                    "dap_parameters must be a dict when lmfit is not installed on the client."
+                )
+            raise TypeError("dap_parameters must be a dict or lmfit.Parameters (or omitted).")
+
+        return Waveform._normalize_param_overrides(parameters)
+
+    @staticmethod
+    def _normalize_param_overrides(parameters: dict) -> dict | None:
+        normalized: dict[str, dict] = {}
+        for name, spec in parameters.items():
+            if spec is None:
+                continue
+            if isinstance(spec, (int, float, np.number)):
+                normalized[name] = {"name": name, "value": float(spec), "vary": False}
+                continue
+            if lmfit is not None and isinstance(spec, lmfit.Parameter):
+                normalized[name] = serialize_param_object(spec)
+                continue
+            if isinstance(spec, dict):
+                normalized[name] = {"name": name, **spec}
+                if "vary" not in normalized[name]:
+                    normalized[name]["vary"] = False
+                continue
+            raise TypeError(
+                f"Invalid dap_parameters entry for '{name}': expected number, dict, or lmfit.Parameter."
+            )
+
+        return normalized or None
+
+    @staticmethod
+    def _format_dap_label(dap_name: str | list[str]) -> str:
+        if isinstance(dap_name, (list, tuple)):
+            return "+".join(dap_name)
+        return dap_name
+
     @SafeSlot(dict, dict)
     def update_dap_curves(self, msg, metadata):
         """
@@ -1793,14 +1936,6 @@ class Waveform(PlotBase):
         if not curve:
             return
 
-        # Get data from the parent (device) curve
-        parent_curve = self._find_curve_by_label(curve.config.parent_label)
-        if parent_curve is None:
-            return
-        x_parent, _ = parent_curve.get_data()
-        if x_parent is None or len(x_parent) == 0:
-            return
-
         # Retrieve and store the fit parameters and summary from the DAP server response
         try:
             curve.dap_params = msg["data"][1]["fit_parameters"]
@@ -1809,19 +1944,13 @@ class Waveform(PlotBase):
             logger.warning(f"Failed to retrieve DAP data for curve '{curve.name()}'")
             return
 
-        # Render model according to the DAP model name and parameters
-        model_name = curve.config.signal.dap
-        model_function = getattr(lmfit.models, model_name)()
-
-        x_min, x_max = x_parent.min(), x_parent.max()
-        oversample = curve.dap_oversample
-        new_x = np.linspace(x_min, x_max, int(len(x_parent) * oversample))
-
-        # Evaluate the model with the provided parameters to generate the y values
-        new_y = model_function.eval(**curve.dap_params, x=new_x)
-
-        # Update the curve with the new data
-        curve.setData(new_x, new_y)
+        # Plot the fitted curve using the server-provided output to avoid requiring lmfit on the client.
+        try:
+            fit_data = msg["data"][0]
+            curve.setData(np.asarray(fit_data["x"]), np.asarray(fit_data["y"]))
+        except Exception as e:
+            logger.exception(f"Failed to plot DAP result for curve '{curve.name()}', error: {e}")
+            return
 
         metadata.update({"curve_id": curve_id})
         self.dap_params_update.emit(curve.dap_params, metadata)
@@ -2341,24 +2470,20 @@ class DemoApp(QMainWindow):  # pragma: no cover
     def __init__(self):
         super().__init__()
         self.setWindowTitle("Waveform Demo")
-        self.resize(1200, 600)
+        self.resize(1600, 600)
         self.main_widget = QWidget(self)
         self.layout = QHBoxLayout(self.main_widget)
         self.setCentralWidget(self.main_widget)
 
-        self.waveform_popup = Waveform(popups=True)
-        self.waveform_popup.plot(device_y="waveform")
-
-        self.waveform_side = Waveform(popups=False)
-        self.waveform_side.plot(device_y="bpm4i", signal_y="bpm4i", dap="GaussianModel")
-        self.waveform_side.plot(device_y="bpm3a", signal_y="bpm3a")
-
         self.custom_waveform = Waveform(popups=True)
         self._populate_custom_curve_demo()
 
-        self.layout.addWidget(self.waveform_side)
-        self.layout.addWidget(self.waveform_popup)
+        self.sine_waveform = Waveform(popups=True)
+        self.sine_waveform.dap_params_update.connect(self._log_sine_dap_params)
+        self._populate_sine_curve_demo()
+
         self.layout.addWidget(self.custom_waveform)
+        self.layout.addWidget(self.sine_waveform)
 
     def _populate_custom_curve_demo(self):
         """
@@ -2377,8 +2502,141 @@ class DemoApp(QMainWindow):  # pragma: no cover
         sigma = 0.8
         y = amplitude * np.exp(-((x - center) ** 2) / (2 * sigma**2)) + noise
 
+        # 1) No explicit parameters: server will use lmfit defaults/guesses.
         self.custom_waveform.plot(x=x, y=y, label="custom-gaussian", dap="GaussianModel")
 
+        # 2) Easy dict: numbers mean "fix this parameter to value" (vary=False).
+        self.custom_waveform.plot(
+            x=x,
+            y=y,
+            label="custom-gaussian-fixed-easy",
+            dap="GaussianModel",
+            dap_parameters={"amplitude": 1.0},
+            dap_oversample=5,
+        )
+
+        # 3) Partial parameter override: this should still trigger guessing on the server
+        #    because not all Gaussian parameters are explicitly specified.
+        self.custom_waveform.plot(
+            x=x,
+            y=y,
+            label="custom-gaussian-partial-guess",
+            dap="GaussianModel",
+            dap_parameters={
+                "center": {"value": 1.2, "vary": True},
+                "sigma": {"value": sigma, "vary": False, "min": 0.0},
+            },
+        )
+
+        # 4) Complete parameter override: this should skip guessing on the server.
+        if lmfit is not None:
+            params_gauss = lmfit.models.GaussianModel().make_params()
+            params_gauss["amplitude"].set(value=amplitude, vary=False)
+            params_gauss["center"].set(value=center, vary=False)
+            params_gauss["sigma"].set(value=sigma, vary=False, min=0.0)
+            self.custom_waveform.plot(
+                x=x,
+                y=y,
+                label="custom-gaussian-complete-no-guess",
+                dap="GaussianModel",
+                dap_parameters=params_gauss,
+            )
+        else:
+            logger.info("Skipping lmfit.Parameters demo (lmfit not installed on client).")
+
+        # Composite example: spectrum with three Gaussians (DAP-only)
+        x_spec = np.linspace(-5, 5, 800)
+        rng_spec = np.random.default_rng(123)
+        centers = [-2.0, 0.6, 2.4]
+        amplitudes = [2.5, 3.2, 1.8]
+        sigmas = [0.35, 0.5, 0.3]
+        y_spec = (
+            amplitudes[0] * np.exp(-((x_spec - centers[0]) ** 2) / (2 * sigmas[0] ** 2))
+            + amplitudes[1] * np.exp(-((x_spec - centers[1]) ** 2) / (2 * sigmas[1] ** 2))
+            + amplitudes[2] * np.exp(-((x_spec - centers[2]) ** 2) / (2 * sigmas[2] ** 2))
+            + rng_spec.normal(loc=0, scale=0.06, size=x_spec.size)
+        )
+
+        # 5) Composite model with partial overrides only: this should still trigger guessing.
+        self.custom_waveform.plot(
+            x=x_spec,
+            y=y_spec,
+            label="custom-gaussian-spectrum-partial-guess",
+            dap=["GaussianModel", "GaussianModel", "GaussianModel"],
+            dap_parameters=[
+                {"center": {"value": centers[0], "vary": False}},
+                {"center": {"value": centers[1], "vary": False}},
+                {"center": {"value": centers[2], "vary": False}},
+            ],
+        )
+
+        # 6) Composite model with all component parameters specified: this should skip guessing.
+        self.custom_waveform.plot(
+            x=x_spec,
+            y=y_spec,
+            label="custom-gaussian-spectrum-complete-no-guess",
+            dap=["GaussianModel", "GaussianModel", "GaussianModel"],
+            dap_parameters=[
+                {
+                    "amplitude": {"value": amplitudes[0], "vary": False},
+                    "center": {"value": centers[0], "vary": False},
+                    "sigma": {"value": sigmas[0], "vary": False, "min": 0.0},
+                },
+                {
+                    "amplitude": {"value": amplitudes[1], "vary": False},
+                    "center": {"value": centers[1], "vary": False},
+                    "sigma": {"value": sigmas[1], "vary": False, "min": 0.0},
+                },
+                {
+                    "amplitude": {"value": amplitudes[2], "vary": False},
+                    "center": {"value": centers[2], "vary": False},
+                    "sigma": {"value": sigmas[2], "vary": False, "min": 0.0},
+                },
+            ],
+        )
+
+    def _populate_sine_curve_demo(self):
+        """
+        Showcase how lmfit's base SineModel can struggle with a drifting baseline.
+        """
+        x = np.linspace(0, 6 * np.pi, 600)
+        rng = np.random.default_rng(7)
+        amplitude = 1.6
+        frequency = 0.75
+        phase = 0.4
+        offset = 0.8
+        slope = 0.08
+        noise = rng.normal(loc=0, scale=0.12, size=x.size)
+        y = offset + slope * x + amplitude * np.sin(2 * np.pi * frequency * x + phase) + noise
+
+        # Base SineModel (no offset support) to show the mismatch
+        self.sine_waveform.plot(x=x, y=y, label="custom-sine-data", dap="SineModel")
+
+        # Composite model: Sine + Linear baseline (offset + slope)
+        self.sine_waveform.plot(
+            x=x,
+            y=y,
+            label="custom-sine-composite",
+            dap=["SineModel", "LinearModel"],
+            dap_oversample=4,
+        )
+
+        if lmfit is None:
+            logger.info("Skipping sine lmfit demo (lmfit not installed on client).")
+            return
+
+        return
+
+    @staticmethod
+    def _log_sine_dap_params(params: dict, metadata: dict):
+        curve_id = metadata.get("curve_id")
+        if curve_id not in {
+            "custom-sine-data-SineModel",
+            "custom-sine-composite-SineModel+LinearModel",
+        }:
+            return
+        logger.info(f"SineModel DAP fit params ({curve_id}): {params}")
+
 
 if __name__ == "__main__":  # pragma: no cover
     import sys

pyproject.toml

@@ -4,7 +4,7 @@ build-backend = "hatchling.build"
 
 [project]
 name = "bec_widgets"
-version = "3.1.4"
+version = "3.2.0"
 description = "BEC Widgets"
 requires-python = ">=3.11"
 classifiers = [

tests/end-2-end/test_bec_dock_rpc_e2e.py

@@ -150,7 +150,9 @@ def test_rpc_gui_obj(connected_client_gui_obj, qtbot):
     # communication should work, main dock area should have same id and be visible
 
     yw = gui.new("Y")
+    qtbot.waitUntil(lambda: len(gui.windows) == 2, timeout=3000)
     yw.delete_all()
     assert len(gui.windows) == 2
     yw.remove()
-    assert len(gui.windows) == 1  # only bec is left
+    qtbot.waitUntil(lambda: len(gui.windows) == 1, timeout=3000)
+    assert len(gui.windows) == 1

tests/end-2-end/test_plotting_framework_e2e.py

@@ -72,6 +72,7 @@ def test_rpc_plotting_shortcuts_init_configs(qtbot, connected_client_gui_obj):
         "dap": None,
         "device": "bpm4i",
         "signal": "bpm4i",
+        "dap_parameters": None,
         "dap_oversample": 1,
     }
     assert c1._config_dict["source"] == "device"

tests/unit_tests/test_waveform.py

@@ -516,6 +516,112 @@ def test_plot_custom_curve_with_inline_dap(qtbot, mocked_client_with_dap):
     assert dap_curve.config.signal.dap == "GaussianModel"
 
 
+def test_normalize_dap_parameters_number_dict():
+    normalized = Waveform._normalize_dap_parameters({"amplitude": 1.0, "center": 2})
+    assert normalized == {
+        "amplitude": {"name": "amplitude", "value": 1.0, "vary": False},
+        "center": {"name": "center", "value": 2.0, "vary": False},
+    }
+
+
+def test_normalize_dap_parameters_dict_spec_defaults_vary_false():
+    normalized = Waveform._normalize_dap_parameters({"sigma": {"value": 0.8, "min": 0.0}})
+    assert normalized["sigma"]["name"] == "sigma"
+    assert normalized["sigma"]["value"] == 0.8
+    assert normalized["sigma"]["min"] == 0.0
+    assert normalized["sigma"]["vary"] is False
+
+
+def test_normalize_dap_parameters_invalid_type_raises():
+    with pytest.raises(TypeError):
+        Waveform._normalize_dap_parameters(["amplitude", 1.0])  # type: ignore[arg-type]
+
+
+def test_normalize_dap_parameters_composite_list():
+    normalized = Waveform._normalize_dap_parameters(
+        [{"center": 1.0}, {"sigma": {"value": 0.5, "min": 0.0}}],
+        dap_name=["GaussianModel", "GaussianModel"],
+    )
+    assert normalized == [
+        {"center": {"name": "center", "value": 1.0, "vary": False}},
+        {"sigma": {"name": "sigma", "value": 0.5, "min": 0.0, "vary": False}},
+    ]
+
+
+def test_normalize_dap_parameters_composite_dict():
+    normalized = Waveform._normalize_dap_parameters(
+        {
+            "GaussianModel": {"center": {"value": 1.0, "vary": True}},
+            "LorentzModel": {"amplitude": 2.0},
+        },
+        dap_name=["GaussianModel", "LorentzModel"],
+    )
+    assert normalized["GaussianModel"]["center"]["value"] == 1.0
+    assert normalized["GaussianModel"]["center"]["vary"] is True
+    assert normalized["LorentzModel"]["amplitude"]["value"] == 2.0
+    assert normalized["LorentzModel"]["amplitude"]["vary"] is False
+
+
+def test_request_dap_includes_normalized_parameters(qtbot, mocked_client_with_dap, monkeypatch):
+    wf = create_widget(qtbot, Waveform, client=mocked_client_with_dap)
+    curve = wf.plot(
+        x=[0, 1, 2],
+        y=[1, 2, 3],
+        label="custom-inline-params",
+        dap="GaussianModel",
+        dap_parameters={"amplitude": 1.0},
+    )
+    dap_curve = wf.get_curve(f"{curve.name()}-GaussianModel")
+    assert dap_curve is not None
+    dap_curve.dap_oversample = 3
+
+    captured = {}
+
+    def capture(topic, msg, *args, **kwargs):  # noqa: ARG001
+        captured["topic"] = topic
+        captured["msg"] = msg
+
+    monkeypatch.setattr(wf.client.connector, "set_and_publish", capture)
+    wf.request_dap()
+
+    msg = captured["msg"]
+    dap_kwargs = msg.content["config"]["kwargs"]
+    assert dap_kwargs["oversample"] == 3
+    assert dap_kwargs["parameters"] == {
+        "amplitude": {"name": "amplitude", "value": 1.0, "vary": False}
+    }
+
+
+def test_request_dap_includes_composite_parameters_list(qtbot, mocked_client_with_dap, monkeypatch):
+    wf = create_widget(qtbot, Waveform, client=mocked_client_with_dap)
+    curve = wf.plot(
+        x=[0, 1, 2],
+        y=[1, 2, 3],
+        label="custom-composite",
+        dap=["GaussianModel", "GaussianModel"],
+        dap_parameters=[{"center": 0.0}, {"center": 1.0}],
+    )
+    dap_curve = wf.get_curve(f"{curve.name()}-GaussianModel+GaussianModel")
+    assert dap_curve is not None
+
+    captured = {}
+
+    def capture(topic, msg, *args, **kwargs):  # noqa: ARG001
+        captured["topic"] = topic
+        captured["msg"] = msg
+
+    monkeypatch.setattr(wf.client.connector, "set_and_publish", capture)
+    wf.request_dap()
+
+    msg = captured["msg"]
+    dap_kwargs = msg.content["config"]["kwargs"]
+    assert dap_kwargs["parameters"] == [
+        {"center": {"name": "center", "value": 0.0, "vary": False}},
+        {"center": {"name": "center", "value": 1.0, "vary": False}},
+    ]
+    assert msg.content["config"]["class_kwargs"]["model"] == ["GaussianModel", "GaussianModel"]
+
+
 def test_fetch_scan_data_and_access(qtbot, mocked_client, monkeypatch):
     """
     Test the _fetch_scan_data_and_access method returns live_data/val if in a live scan,