RubarBE: split critical Shields stress (TCADIM) from water viscosity (VISC)

src/Model/SolverParameters/SolverParametersList.py

@@ -315,6 +315,14 @@ class SolverParametersList(PamhyrModelList):
 
             n._lst = list(map(lambda t: t[1], sorted(lst)))
 
+            # Parameters added to the solver after the study was saved
+            names = {p.name for p in n._lst}
+            n._lst += [
+                Parameter(name=name, value=value, status=status)
+                for name, value in st.default_parameters()
+                if name not in names
+            ]
+
             new[solver] = n
 
         return new

src/Scripts/Run.py

@@ -25,6 +25,7 @@ from PyQt5.QtCore import QProcess
 
 from Scripts.AScript import AScript
 from Model.Study import Study
+from View.RunSolver.Translate import RunSolverTranslate
 
 logger = logging.getLogger()
 
@@ -89,12 +90,16 @@ class ScriptRun(AScript):
 
         # Export and Run
         logger.info(f"~Export~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
-        solver.export(study, workdir, qlog=self._q)
+        ok = solver.export(study, workdir, qlog=self._q)
 
         while self._q.qsize() != 0:
             s = self._q.get()
             logger.info(s)
 
+        if not ok:
+            logger.warning(RunSolverTranslate()["export_failed"])
+            return
+
         if command == "run":
             logger.info(f"~Run~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~")
             solver.run(

src/Solver/RubarBE.py

@@ -18,11 +18,14 @@
 
 import os
 import logging
+import math
+from decimal import Decimal, InvalidOperation, localcontext
 import numpy as np
 
 from tools import timer, trace, old_pamhyr_date_to_timestamp, logger_exception
 
 from Solver.CommandLine import CommandLineSolver
+from View.RunSolver.Translate import RunSolverTranslate
 
 from Model.Results.Results import Results
 from Model.Results.River.River import River, Reach, Profile
@@ -44,7 +47,7 @@ class Rubar3(CommandLineSolver):
 
     @classmethod
     def default_parameters(cls):
-        # lst = super(Rubar3, cls).default_parameters()
+        # Keep historical rubarbe_* keys for existing Rubar3 studies.
 
         lst = [
             ("rubarbe_cfl", "0.50000E+00"),
@@ -52,13 +55,13 @@ class Rubar3(CommandLineSolver):
             ("rubarbe_condav", "3"),
             ("rubarbe_regime", "0"),
             ("rubarbe_iodev", "n"),
-            ("rubarbe_iodebord", ""),
+            ("rubarbe_iodebord", "o"),
-            ("rubarbe_iostockage", ""),
+            ("rubarbe_iostockage", "o"),
             ("rubarbe_iopdt", "y"),
             ("rubarbe_iovis", "n"),
             ("rubarbe_rep", "n"),
             ("rubarbe_tinit", "000:00:00:00"),
-            ("rubarbe_tmax", "000:01:00:00"),  # 1 day
+            ("rubarbe_tmax", "001:00:00:00"),  # 1 day
             ("rubarbe_tiopdt", "000:00:00:00"),
             ("rubarbe_dt", "5.0"),
             ("rubarbe_ts", "999:99:99:00"),
@@ -89,8 +92,9 @@ class Rubar3(CommandLineSolver):
             ("rubarbe_sediment_mult_3", ""),
             ("rubarbe_sediment_mult_4", ""),
             ("rubarbe_sediment_mult_5", ""),
-            ("rubarbe_sediment_visc", "0.047"),
+            ("rubarbe_sediment_tcadim", "0.047"),
-            ("rubarbe_sediment_opts", "6"),
+            ("rubarbe_sediment_visc", "1.3E-06"),
+            ("rubarbe_sediment_opts", "3"),
             ("rubarbe_sediment_odchar", "0"),
             ("rubarbe_sediment_unisol", "1"),
             ("rubarbe_sediment_typdef", "3"),
@@ -140,8 +144,34 @@ class Rubar3(CommandLineSolver):
         self._study = study
         name = study.name.replace(" ", "_")
 
+        values = self._parameter_values(study)
+        if (values.get("rubarbe_regime") == "1" and
+                values.get("rubarbe_condam") == "0"):
+            trad = RunSolverTranslate()
+            message = trad["incompatible_upstream_boundary"]
+            logger.warning(message)
+            if qlog is not None:
+                qlog.put(trad["warning"].format(message=message))
+            return False
+
+        trased = self._parameter_flag(study, "trased") == "O"
+        if trased:
+            # Notice §3.6: HYDROS is required at each upstream boundary
+            missing = self._reaches_without_solid_boundary(study)
+            if missing:
+                trad = RunSolverTranslate()
+                message = trad["missing_solid_boundary"].format(
+                    reaches=", ".join(missing)
+                )
+                logger.warning(message)
+                if qlog is not None:
+                    qlog.put(trad["warning"].format(message=message))
+                return False
+
         self._export_donnee(study, repertory, qlog, name=name)
-        self._export_ts(study, repertory, qlog, name=name)
+        if trased:
+            if not self._export_ts(study, repertory, qlog, name=name):
+                return False
         self._export_geomac_i(study, repertory, qlog, name=name)
         self._export_mail(study, repertory, qlog, name=name)
         self._export_devers(study, repertory, qlog, name=name)
@@ -153,108 +183,247 @@ class Rubar3(CommandLineSolver):
 
         return True
 
-    def _export_donnee(self, study, repertory, qlog, name="0"):
+    # Notice Rubar3/RubarBE, annex 1: records are positional, not keyed.
+    # Each record starts with an A50 label, including records of integers.
+    _donnee_fields = (
+        ("cfl", "real"),
+        ("condam", "integer"), ("condav", "integer"),
+        ("regime", "integer"), ("iodev", "flags"),
+        ("iopdt", "flag"), ("iovis", "flag"), ("rep", "flag"),
+        ("tinit", "time"), ("tmax", "time"), ("tiopdt", "time"),
+        ("dt", "real"), ("ts", "time"), ("dtsauv", "time"),
+        ("psave", "time"), ("fdeb1", "real"), ("fdeb2", "real"),
+        ("fdeb3", "real"),
+        ("tf_1", "flag"), ("tf_2", "flag"), ("tf_3", "flag"),
+        ("tf_4", "flag"), ("tf_5", "flag"), ("tf_6", "flag"),
+        ("trased", "flag"),
+    )
+    _ts_fields = (
+        ("ros", "real"), ("por", "real"), ("dcharg", "real"),
+        ("halfa", "real"), ("mult_1", "multipliers"),
+        ("visc", "visc"), ("opts", "law"), ("odchar", "integer"),
+        ("unisol", "integer"), ("typdef", "integer"),
+        ("depot", "depot"), ("choixc", "integer"),
+        ("option", "options"), ("capsol", "integer"),
+        ("bmiu", "real"), ("demix", "integer"),
+        ("defond", "integer"), ("varcons", "integer"),
+    )
+
+    # rubar_init.f90 (inits): laws whose critical stress may come from the
+    # Shields diagram, and laws without critical stress
+    _shields_laws = {1, 10, 11, 12, 14, 25, 26}
+    _no_threshold_laws = {2, 3, 8, 13, 18, 19, 22, 23, 24}
+    _brownlie_law = 16
+
+    @classmethod
+    def tcadim_required(cls, opts, option):
+        """True if RubarBE reads the dimensionless critical stress TCADIM
+        for transport law OPTS and options OPTION (notice §3.10)"""
+        try:
+            law = int(opts)
+        except ValueError:
+            return False
+
+        shields = len(option) == 3 and option[1] != "1"
+        if shields and law in cls._shields_laws:
+            return False
+        if law in cls._no_threshold_laws or law == cls._brownlie_law:
+            return False
+        return True
+
+    @staticmethod
+    def _reaches_without_solid_boundary(study):
+        river = study.river
+        solid = [
+            bc for bc in river.boundary_condition.get_tab("solid")
+            if not bc.is_deleted() and bc.bctype == "SL" and len(bc) > 0
+        ]
+
+        return [
+            edge.name for edge in river.enable_edges()
+            if river.is_upstream_node(edge.node1)
+            and not any(bc.node == edge.node1 for bc in solid)
+        ]
+
+    def _parameter_values(self, study):
+        return {p.name: str(p.value).strip()
+                for p in study.river.get_params(self.type).parameters}
+
+    @staticmethod
+    def _required(values, key):
+        value = values.get(key, "")
+        if not value:
+            raise ValueError(f"Missing Rubar parameter: {key}")
+        return value
+
+    @staticmethod
+    def _flag(value, key, mixed=False):
+        result = value.upper()
+        if result == "Y":
+            result = "O"
+        if result not in (("O", "N", "M") if mixed else ("O", "N")):
+            raise ValueError(f"{key}: expected O/Y or N" +
+                             (" or M" if mixed else ""))
+        return result
+
+    def _parameter_flag(self, study, suffix):
+        key = "rubarbe_" + suffix
+        return self._flag(self._required(self._parameter_values(study), key),
+                          key)
+
+    @staticmethod
+    def _integer(value, key, width=1):
+        if not value.isascii() or not value.isdigit() or len(value) > width:
+            raise ValueError(f"{key}: expected an integer in {width} columns")
+        return value.rjust(width)
+
+    @staticmethod
+    def _real(value, key, scientific=False):
+        try:
+            number = Decimal(value.replace("d", "E").replace("D", "E"))
+        except InvalidOperation:
+            raise ValueError(f"{key}: invalid real value {value!r}") from None
+        if not number.is_finite():
+            raise ValueError(f"{key}: expected a finite real value")
+        if scientific:
+            # DONNEE: E12.5. Reject overflow instead of shifting columns.
+            floating = float(number)
+            if not math.isfinite(floating) or (number and floating == 0):
+                raise ValueError(f"{key}: real value outside E12.5 range")
+            result = f"{floating:12.5E}"
+            width = 12
+        else:
+            # TS: F10.0 input accepts an explicit decimal point/exponent.
+            # Preserve the value exactly; do not round porosity etc. to int.
+            with localcontext() as context:
+                context.prec = max(28, len(number.as_tuple().digits))
+                number = number.normalize()
+            result = str(number).upper()
+            if len(result) > 10:
+                result = format(number, "E")
+            width = 10
+        if len(result) > width:
+            raise ValueError(f"{key}: value {value!r} exceeds {width} columns")
+        return result.rjust(width)
+
+    @staticmethod
+    def _write_parameters(repertory, filename, lines):
+        # Validate all records before opening/truncating the destination.
+        path = os.path.join(repertory, filename)
+        with open(path, "w", encoding="ascii") as f:
+            f.writelines(lines)
+
+    def _export_donnee(self, study, repertory, qlog=None, name="0"):
+        values = self._parameter_values(study)
+        fields = list(self._donnee_fields)
+        # The diagnostic transport-capacity extension is optional as a block.
+        if values.get("rubarbe_optfpc", ""):
+            fields += [("optfpc", "integer"), ("ros", "real"),
+                       ("dm", "real"), ("segma", "real")]
+        elif any(values.get("rubarbe_" + key, "")
+                 for key in ("ros", "dm", "segma")):
+            raise ValueError("rubarbe_optfpc: required before ROS/DM/SEGMA")
+        lines = []
+        for suffix, kind in fields:
+            key = "rubarbe_" + suffix
+            value = self._required(values, key)
+            if kind == "integer":
+                result = self._integer(value, key)
+            elif kind in ("flag", "flags"):
+                result = self._flag(value, key, mixed=suffix == "iopdt")
+                if kind == "flags":
+                    # Missing optional flags mean O (notice section 3.1).
+                    # Write both explicitly so STOCKAGE cannot shift left.
+                    for extra in ("iodebord", "iostockage"):
+                        extra_key = "rubarbe_" + extra
+                        result += self._flag(values.get(extra_key) or "O",
+                                             extra_key)
+            else:
+                if kind == "time" and ":" in value:
+                    parts = value.split(":")
+                    if len(parts) != 4 or not all(p.isdigit() for p in parts):
+                        raise ValueError(
+                            f"{key}: expected days:hours:minutes:seconds")
+                    value = str(old_pamhyr_date_to_timestamp(value))
+                result = self._real(value, key, scientific=True)
+            lines.append(f"{key:<50}{result}\n")
         if qlog is not None:
             qlog.put("Export DONNEE file")
+        self._write_parameters(repertory, f"donnee.{name}", lines)
 
-        with open(
+    def _export_ts(self, study, repertory, qlog=None, name="0"):
-                os.path.join(
+        trad = RunSolverTranslate()
-                    repertory, f"donnee.{name}"
+        values = self._parameter_values(study)
-                ), "w+"
+        prefix = "rubarbe_sediment_"
-        ) as f:
+        fields = list(self._ts_fields)
-            params = filter(
+        demix = self._required(values, prefix + "demix")
-                lambda p: "rubarbe_sediment_" not in p.name,
+        if demix not in ("0", "1"):
-                study.river.get_params(self._type).parameters
+            raise ValueError(f"{prefix}demix: expected 0 or 1")
-            )
+        if demix == "1":
-
+            fields += [("dchard", "real"), ("dchars", "real")]
-            it = iter(params)
+        lines = []
-
+        for suffix, kind in fields:
-            param = next(it, None)
+            key = prefix + suffix
-            while param is not None:
+            value = self._required(values, key)
-                name = param.name
+            if kind == "real":
-                value = param.value
+                result = self._real(value, key)
-
+            elif kind == "visc":
-                if "all_" in name:
+                # Notice annex 1: TCADIM then VISC, read as (A50, A20) then
-                    continue
+                # two F10.0. A single value would be read as TCADIM, with
-
+                # VISC = 0 (rubar_init.f90, inits).
-                if value != "":
+                if Decimal(self._real(value, key)) <= 0:
-                    # Value format
+                    raise ValueError(f"{key}: expected a positive viscosity")
-                    if value.count(':') == 3:
+                tcadim_key = prefix + "tcadim"
-                        value = old_pamhyr_date_to_timestamp(value)
+                tcadim = values.get(tcadim_key, "")
-                        value = f"{value:>12.5e}".upper()
+                if tcadim:
-
+                    result = self._real(tcadim, tcadim_key)
-                    if value.count('.') == 1:
+                elif self.tcadim_required(values.get(prefix + "opts", ""),
-                        value = f"{float(value):>12.5e}".upper()
+                                          values.get(prefix + "option", "")):
-
+                    message = trad["missing_tcadim"].format(key=tcadim_key)
-                    if value == "y" or value == "n":
+                    logger.warning(message)
-                        value = "O" if value == "y" else "N"
+                    if qlog is not None:
-
+                        qlog.put(trad["warning"].format(message=message))
-                    # Write value
+                    return False
-                    f.write(f"{name:<50}{value}")
+                else:
-
+                    result = " " * 10
-                    # Add values of 'rubarbe_iodebord' and
+                result += self._real(value, key)
-                    # 'rubarbe_iostockage'
+            elif kind == "multipliers":
-                    if name == "rubarbe_iodev":
+                result = self._real(value, key)
-                        v2 = next(it).value
+                extras = [values.get(prefix + f"mult_{i}", "")
-                        v3 = next(it).value
+                          for i in range(2, 6)]
-
+                # Preserve empty interior F10.0 fields; omit trailing fields
-                        # f.write(f"{v2}{v3}")
+                # to retain the solver's defaults for optional coefficients.
-
+                while extras and not extras[-1]:
-                    # New line
+                    extras.pop()
-                    f.write(f"\n")
+                for i, extra in enumerate(extras, 2):
-
+                    result += (self._real(extra, prefix + f"mult_{i}")
-                param = next(it, None)
+                               if extra else " " * 10)
-
+            elif kind == "options":
-    def _export_ts(self, study, repertory, qlog, name="0"):
+                if len(value) != 3 or any(c not in "12" for c in value):
+                    message = trad["invalid_options"].format(key=key)
+                    logger.warning(message)
+                    if qlog is not None:
+                        qlog.put(trad["warning"].format(message=message))
+                    return False
+                result = value
+            elif kind == "law":
+                # Notice section 3.10 / 8.7: 4--7 are not listed.
+                if value not in {str(i) for i in [1, 2, 3, *range(8, 31)]}:
+                    message = trad["unsupported_transport_law"].format(
+                        key=key, value=value)
+                    logger.warning(message)
+                    if qlog is not None:
+                        qlog.put(trad["warning"].format(message=message))
+                    return False
+                result = self._integer(value, key, width=2)
+            else:
+                result = self._integer(value, key,
+                                       width=10 if kind == "depot" else 1)
+            lines.append(f"{key:<50}{result}\n")
         if qlog is not None:
             qlog.put("Export TS file")
-
+        self._write_parameters(repertory, f"ts.{name}", lines)
-        with open(
+        return True
-                os.path.join(
-                    repertory, f"ts.{name}"
-                ), "w+"
-        ) as f:
-            def float_format(string):
-                if "." in string:
-                    return f"{float(string):>10.0f}"
-                return ""
-
-            params = filter(
-                lambda p: "rubarbe_sediment_" in p.name,
-                study.river.get_params(self.type).parameters
-            )
-            it = iter(params)
-
-            line = 0
-            while line < 20:
-                param = next(it)
-                name = param.name
-                value = param.value
-
-                if value != "":
-                    # Value format
-                    if value.count('.') == 1:
-                        value = f"{float_format(value)}"
-                    else:
-                        value = f"{value:>10}"
-
-                    # Write value
-                    f.write(f"{name:<50}{value}")
-
-                    # Add values of 'rubarbe_iodebord' and
-                    # 'rubarbe_iostockage'
-                    if name == "rubarbe_sediment_mult_1":
-                        m2 = f"{float_format(next(it).value)}"
-                        m3 = f"{float_format(next(it).value)}"
-                        m4 = f"{float_format(next(it).value)}"
-                        m5 = f"{float_format(next(it).value)}"
-
-                        f.write(f"{m2}{m3}{m4}{m5}")
-
-                    # New line
-                    f.write(f"\n")
-
-                line += 1
 
     def _export_geomac_i(self, study, repertory, qlog, name="0"):
         if qlog is not None:
@@ -616,6 +785,7 @@ class Rubar3(CommandLineSolver):
             )).value
             if timestamp.count(':') == 3:
                 timestamp = old_pamhyr_date_to_timestamp(timestamp)
+            timestamp = float(timestamp)
             ts.add(timestamp)
 
             # add initial condition
@@ -636,8 +806,7 @@ class Rubar3(CommandLineSolver):
             while True:
                 line = f.readline()
                 if line == "":
-                    results.set("timestamps", ts)
+                    break
-                    return
 
                 timestamp = float(line)
                 ts.add(timestamp)
@@ -671,6 +840,24 @@ class Rubar3(CommandLineSolver):
                     h, s, q, z = read_data_line(f)
                     set_and_compute_limites(reach, ind, z+h, q, s)
 
+            # Tables and plots use timestamp rows and global profile columns.
+            timestamps = sorted(ts)
+            profiles = []
+            for reach, _ in reachs:
+                offset = len(profiles)
+                profiles.extend(reach.profiles)
+                reach.set_global_index(range(offset, len(profiles)))
+
+            results.set("timestamps", timestamps)
+            results.compute_timestamp_index()
+            results.set("table", {
+                key: np.array([
+                    [profile.get_ts_key(t, key) for profile in profiles]
+                    for t in timestamps
+                ], dtype=float)
+                for key in ("Z", "Q", "V")
+            })
+
     @timer
     def write_bin(self, study, fname, results, qlog=None, name="0"):
         logger.info(f"write_bin: Start writing '{fname}' ...")
@@ -800,3 +987,14 @@ class RubarBE(Rubar3):
         self._cmd_input = ""
         self._cmd_solver = "@path @args @input"
         self._cmd_output = ""
+
+    @classmethod
+    def default_parameters(cls):
+        # New BE studies evolve the bed. Keep Rubar3 defaults unchanged.
+        overrides = {
+            "rubarbe_trased": "y",
+            "rubarbe_sediment_opts": "1",  # Meyer-Peter and Mueller
+            "rubarbe_sediment_option": "212",  # no Ikeda, given Shields, total
+        }
+        return [(key, overrides.get(key, value))
+                for key, value in super().default_parameters()]

src/Solver/Solvers.py

@@ -35,7 +35,7 @@ solver_long_name = {
     # "mage_fake7": "Mage fake v7",
     "adistswc": "Adis-TS_WC",
     "adisttwc": "Adis-TT_WC",
-    # "rubarbe": "RubarBE",
+    "rubarbe": "RubarBE",
     "rubar3": "Rubar3",
 }
 
@@ -46,6 +46,6 @@ solver_type_list = {
     # "mage_fake7": MageFake7,
     "adistswc": AdisTSwc,
     "adisttwc": AdisTTwc,
-    # "rubarbe": RubarBE,
+    "rubarbe": RubarBE,
     "rubar3": Rubar3,
 }

src/View/RunSolver/Translate.py

@@ -0,0 +1,61 @@
+# Translate.py -- Pamhyr
+# Copyright (C) 2023-2026  INRAE
+#
+# This program is free software: you can redistribute it and/or modify
+# it under the terms of the GNU General Public License as published by
+# the Free Software Foundation, either version 3 of the License, or
+# (at your option) any later version.
+#
+# This program is distributed in the hope that it will be useful,
+# but WITHOUT ANY WARRANTY; without even the implied warranty of
+# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+# GNU General Public License for more details.
+#
+# You should have received a copy of the GNU General Public License
+# along with this program.  If not, see <https://www.gnu.org/licenses/>.
+
+# -*- coding: utf-8 -*-
+
+from PyQt5.QtCore import QCoreApplication
+
+from View.Tools.PamhyrTranslate import PamhyrTranslate
+
+_translate = QCoreApplication.translate
+
+
+class RunSolverTranslate(PamhyrTranslate):
+    def __init__(self):
+        super(RunSolverTranslate, self).__init__()
+
+        self._dict["invalid_options"] = _translate(
+            "Solver",
+            "Export cancelled: {key}: expected three digits, each 1 or 2, "
+            "Select valid options in the parameters (tab Solid Transport)."
+        )
+        self._dict["unsupported_transport_law"] = _translate(
+            "Solver",
+            "Export cancelled: {key}: unsupported transport law {value!r}. "
+            "Please choose another transport law from the list."
+        )
+        self._dict["warning"] = _translate("Solver", "Warning: {message}")
+        self._dict["incompatible_upstream_boundary"] = _translate(
+            "Solver",
+            "Export cancelled: REGIME=1 is incompatible with CONDAM=0. "
+            "A reflecting upstream boundary cannot impose the water depth."
+        )
+        self._dict["missing_tcadim"] = _translate(
+            "Solver",
+            "Export cancelled: {key}: the selected transport law and options "
+            "use a dimensionless critical stress given by the user. "
+            "Fill it in the parameters (tab Solid Transport)."
+        )
+        self._dict["missing_solid_boundary"] = _translate(
+            "Solver",
+            "Export cancelled: sediment transport (TRASED) is enabled but "
+            "no solid boundary condition is defined at the upstream node "
+            "of reach(es): {reaches}. Add a solid boundary condition, or "
+            "disable sediment transport in the solver parameters."
+        )
+        self._dict["export_failed"] = _translate(
+            "Solver", "Export failed; task cancelled"
+        )

src/View/SolverParameters/RubarFields.py

@@ -0,0 +1,355 @@
+"""Rubar parameter descriptions (notice sections 3.1, 3.10 and annex 1).
+
+Keys and values keep their historical database representation.
+"""
+
+from dataclasses import dataclass
+from decimal import Decimal, DecimalException, InvalidOperation
+
+from Solver.RubarBE import Rubar3
+
+
+GROUPS = ('Temps et calcul', 'Hydraulique', 'Transport solide',
+          'Évolution du fond', 'Sauvegardes')
+YES_NO = {'y': 'Oui', 'n': 'Non'}
+ON_OFF = {'1': 'Oui', '0': 'Non'}
+LAWS = {
+    '1': 'Meyer-Peter et Müller', '2': 'Engelund et Hansen', '3': 'Bagnold',
+    '8': 'Ackers et White (1973)', '9': 'Smart et Jaeggi', '10': 'van Rijn',
+    '11': 'Rickenmann', '12': 'Camenen et Larson', '13': 'Schoklitsch',
+    '14': 'Sato, Hikkawa et Ashida', '15': 'Recking et al. (2008)',
+    '16': 'Brownlie', '17': 'Laursen', '18': 'Ackers et White (1990)',
+    '19': 'Karim', '20': 'Karim et Kennedy', '21': 'Yang et Lim (2003)',
+    '22': 'Yang (1984)', '23': 'Yang (1979)', '24': 'Yang (1973)',
+    '25': 'van Rijn (1984) — suspension', '26': 'van Rijn (1984) — charriage',
+    '27': 'Recking et al. (2016)', '28': 'Piton et al. (2017)',
+    '29': 'Lefort (2007)', '30': 'Lefort (2015)',
+}
+OPTION_CHOICES = (
+    ('Correction de pente',
+     {'1': 'Correction d’Ikeda', '2': 'Sans correction'},
+     'Permet d\'introduire que la répartition de la contrainte critique dans '
+     'une section en travers est influencée par la pente transversale'),
+    ('Seuil critique', {'1': 'Fourni par l’utilisateur',
+                        '2': 'Calculé par le diagramme de Shields'},
+     'Permet de choisir pour certaines formules de capacité, entre un seuil '
+     'critique fourni par l’utilisateur ou '
+     'un seuil critique calculé par le diagramme de Shields.'),
+    ('Contrainte utilisée', {'1': 'Contrainte efficace',
+                             '2': 'Contrainte non réduite'},
+     'Permet de choisir pour certaines formules de capacité, entre la '
+     'contrainte efficace calculée ou la contrainte non réduite.'),
+)
+
+
+@dataclass(frozen=True)
+class Field:
+    key: str
+    group: int
+    label: str
+    help: str
+    kind: str = 'real'
+    choices: dict = None
+    minimum: float = 0
+    maximum: float = None
+    positive: bool = False
+    optional: bool = False
+
+
+def fields():
+    """Describe historical Rubar fields without changing their defaults."""
+    result = []
+
+    def add(key, group, label, help, **kwargs):
+        result.append(Field('rubarbe_' + key, group, label, help, **kwargs))
+
+    def choice(key, group, label, help, choices):
+        add(key, group, label, help, kind='choice', choices=choices)
+
+    add('cfl', 0, 'Nombre de Courant (CFL)',
+        'Limite supérieure du nombre de Courant, afin d\'assurer la stabilité '
+        'du schéma numérique ; comprise entre 0 et 1.',
+        positive=True, maximum=1)
+    for key, label, help in (
+        ('tinit', 'Début du calcul',
+         'Doit correspondre aux conditions initiales.'),
+        ('tmax', 'Fin du calcul', 'Doit être postérieure au début du calcul.'),
+    ):
+        add(key, 0, label + ' (j:h:min:s ou s)', help, kind='time')
+    choice('iopdt', 0, 'Gestion du pas de temps',
+           'En mode mixte, le pas reste fixe jusqu’à la date de bascule.',
+           {'y': 'Automatique', 'n': 'Fixe',
+            'm': 'Mixte : fixe puis automatique'})
+    add('dt', 0, 'Pas de temps initial ou fixe (s)',
+        'Premier pas en mode automatique, pas demandé en mode fixe.',
+        positive=True)
+    add('tiopdt', 0, 'Fin de la période à pas fixe' + ' (j:h:min:s ou s)',
+        'Utilisée uniquement avec un pas de temps mixte.', kind='time')
+    choice('rep', 0, 'Reprendre un calcul',
+           'Nécessite les fichiers de reprise du calcul précédent.', YES_NO)
+
+    choice('condam', 1, 'Condition à la limite amont',
+           'Le choix doit correspondre aux données de la limite amont.',
+           {'0': 'Réflexion', '1': 'Débit imposé Q(t)',
+            '2': 'Cote imposée Z(t)', '4': 'Confluence'})
+    choice('condav', 1, 'Condition à la limite aval',
+           'Le choix doit correspondre aux données de la limite aval.',
+           {'0': 'Sortie libre', '1': 'Courbe de tarage Q(Z)',
+            '2': 'Réflexion',
+            '3': 'Cote imposée Z(t)', '4': 'Confluence / défluence',
+            '5': 'Régime uniforme', '7': 'Régime critique'})
+    choice('regime', 1, 'Imposer aussi le tirant d’eau amont',
+           'Complète le débit imposé ; incompatible avec une limite '
+           'amont de type "Réflexion".',
+           ON_OFF)
+    choice('iodev', 1, 'Déversement latéral',
+           'Nécessite les données de déversements latéraux.',
+           YES_NO)
+    for key, label, help in (
+        ('iodebord', 'Formule Debord',
+         'Oui: utilise la formule Debord pour les lits composés.\nNon:'
+         ' pas de perte de charge complémentaire entre lit mineur et moyen.'),
+        ('iostockage', 'Stockage dans le lit majeur',
+         'Traite le lit majeur sous la berge la plus basse comme zone '
+         'de stockage.'),
+    ):
+        choice(key, 1, label, help, YES_NO)
+    choice('optfpc', 1, 'Calcul du frottement',
+           'Strickler fourni ou frottement calculé à partir des sédiments.',
+           {'0': 'Coefficients de Strickler fournis', '1': 'Brownlie (1983)',
+            '2': 'Griffith (1981) - pour un fond mobile', '3': 'Karim (1995)',
+            '4': 'Wu et Wang (1999)', '5': 'Yu et Lim (2003)',
+            '6': 'Recking et al. (2008)', '7': 'Garde et Raju (1966)'})
+    for key, label in (
+        ('ros', 'Masse volumique des sédiments (kg/m³)'),
+        ('dm', 'Diamètre médian des sédiments (m)'),
+        ('segma', 'Étendue granulométrique des sédiments (—)'),
+    ):
+        add(key, 1, label,
+            'Utilisé pour le frottement calculé sans transport solide '
+            '(disponible uniquement si le transport solide est désactivé '
+            'et que le calcul du frottement se fait avec les sédiments).',
+            positive=True)
+
+    choice('trased', 2, 'Activer le transport solide',
+           'Fonction de RubarBE. L’évolution du fond se règle séparément.',
+           YES_NO)
+    add('sediment_ros', 2, 'Masse volumique des sédiments (kg/m³)',
+        'Masse volumique des grains.', positive=True)
+    add('sediment_por', 2, 'Porosité des sédiments (—)',
+        'Fraction du volume occupée par les vides : de 0 inclus à 1 exclu.',
+        positive=True, maximum=1)
+    choice('sediment_opts', 2, 'Loi de capacité solide',
+           'Formule de capacité solide ; choisir selon le domaine '
+           'physique.',
+           LAWS)
+    choice('sediment_unisol', 2, 'Unité des débits ou concentrations solides',
+           'Unité des débits ou concentrations solides.',
+           {'1': 'Débit massique (kg/s)', '2': 'Concentration (kg/m³)'})
+    choice('sediment_odchar', 2, 'Distance de chargement',
+           'Option de calcul de la ditance de chargement.',
+           {'0': 'Distance constante', '1': 'Loi de Han', '2': 'Wu et Wang'})
+    add('sediment_dcharg', 2, 'Distance de chargement constante (m)',
+        'Utilisée uniquement si la distance de chargement est constante.')
+    add('sediment_halfa', 2, 'Paramètre alpha de Han (m)',
+        'Utilisé uniquement avec la loi de chargement de Han.')
+    add('sediment_option', 2, 'Calcul de la contrainte critique',
+        'Trois choix indépendants, enregistrés ensemble dans OPTION.',
+        kind='options')
+    add('sediment_tcadim', 2, 'Seuil critique adimensionnel (—)',
+        'Contrainte critique adimensionnelle (TCADIM) fournie par '
+        'l’utilisateur, par exemple 0,047 pour Meyer-Peter et Müller ou 0,15 '
+        'pour Recking et al. (2008). Utilisée lorsque le seuil n’est pas '
+        'calculé par le diagramme de Shields et que la loi en a besoin.',
+        positive=True, optional=True)
+    add('sediment_visc', 2, 'Viscosité cinématique de l’eau (m²/s)',
+        'Viscosité cinématique de l’eau, variable avec la température. '
+        'À 15 °C : environ 1,30 × 10⁻⁶ m²/s.', positive=True)
+    choice('sediment_choixc', 2, 'Calcul de la contrainte hydrodynamique de '
+           'cisaillement',
+           'MPC : méthode des perpendiculaires confondues (notice §8.5).',
+           {'1': 'MPC', '2': 'Répartition uniforme',
+            '3': 'MPC — pente moyenne sur 3 segments',
+            '4': 'MPC — moyenne sur 3 segments',
+            '5': 'MPC — moyenne sur une distance de périmètre de '
+                 'longueur égale à la hauteur d’eau',
+            '6': 'MPC — proportionnelle au diamètre de la couche '
+                 'puissance 1/3',
+            '7': 'MPC — proportionnelle au Strickler puissance −3/2',
+            '8': 'MPC — proportionnelle à la hauteur d’eau'})
+    choice('sediment_capsol', 2, 'Calcul de la capacité solide',
+           'Les méthodes 2 et 3 nécessitent une répartition MPC.',
+           {'1': 'Contrainte moyenne, largeur active imposée',
+            '2': 'Contrainte moyenne, largeur active calculée',
+            '3': 'Répartition locale des contraintes'})
+    add('sediment_mult_1', 2, 'Multiplicateur de la capacité de chariage (—)',
+        '1 conserve la capacité calculée par la loi choisie.')
+
+    choice('sediment_defond', 3, 'Faire évoluer la géométrie du lit',
+           'Le transport solide peut être calculé sans déformation du lit.',
+           ON_OFF)
+    choice('sediment_demix', 3, 'Activer le démixage',
+           'Option pour le demixage de deux compartiments sédimentaires.',
+           ON_OFF)
+    for key, label in (
+        ('dchard', 'Distance de chargement pour le diamètre de sédiment (m)'),
+        ('dchars', 'Distance de chargement de l’étendue granulométrique des '
+                   'sédiments (m)'),
+    ):
+        add('sediment_' + key, 3, label,
+            'Utilisée uniquement lorsque le démixage est activé.')
+    choice('sediment_typdef', 3, 'Actualisation de la masse active',
+           'Actualise la masse active après une variation relative de 10⁻ⁿ.',
+           {'0': 'À chaque pas', '1': 'Variation de 10 %',
+            '2': 'Variation de 1 %', '3': 'Variation de 0,1 %',
+            '4': 'Variation de 0,01 %'})
+    choice('sediment_depot', 3, 'Méthode d\'application du dépôt',
+           'Méthodes de déformation de la section décrites au §8.5.3.',
+           {'1': 'Couches horizontales',
+            '2': 'Uniforme sur le périmètre mouillé',
+            '3': 'Inverse de la contrainte locale (1/τ)',
+            '4': 'Selon le seuil critique — tout le lit actif mouillé',
+            '5': 'Selon le seuil critique — une partie du lit actif mouillé',
+            '6': 'Méthode 6 — fonction des contraintes (notice §8.5.3)',
+            '7': 'Méthode 7 — fonction des contraintes (notice §8.5.3)',
+            '8': 'Méthode 8 — code cité dans l’annexe 1'})
+    choice('sediment_varcons', 3, 'Grandeurs conservées après déformation',
+           'Les autres grandeurs hydrauliques sont recalculées.',
+           {'1': 'Section mouillée et débit', '2': 'Niveau d’eau et débit',
+            '3': 'Niveau d’eau et vitesse'})
+    add('sediment_bmiu', 3, 'Tangente de l’angle de frottement interne (—)',
+        'Saisir tan(angle), et non l’angle en degrés, de la valeur moyenne '
+        'de la tangente de l\'angle de frottement interne de sédiments.',
+        positive=True)
+    for i, label in enumerate((
+        'Épaisseur de couche active : multiplicateur du d84',
+        'Épaisseur de fines infiltrantes : multiplicateur du d16',
+        'Coefficient de mélange des couches superficielles',
+        'Coefficient de création d’une couche de dépôt',
+    ), 2):
+        add(f'sediment_mult_{i}', 3, label + ' (—)',
+            'Facultatif : laisser vide pour utiliser le comportement '
+            'du solveur.',
+            optional=True)
+
+    for key, label, help in (
+        ('ts', 'Instant de sauvegarde pour reprise',
+         'Écrit les fichiers de reprise lorsque cet instant est dépassé.'),
+        ('dtsauv', 'Intervalle de sauvegarde des profils',
+         '0 demande une sauvegarde à chaque pas de calcul.'),
+        ('psave', 'Intervalle de sauvegarde des hydrogrammes',
+         '0 demande une sauvegarde à chaque pas de calcul.'),
+    ):
+        add(key, 4, label + ' (j:h:min:s ou s)', help, kind='time')
+    for i, label in enumerate((
+        'Fichiers de reprise', 'Hydrogrammes aux abscisses choisies',
+        'Enveloppes des résultats', 'Profils aux instants choisis',
+        'Profils à intervalle fixe', 'Temps de dépassement des débits',
+    ), 1):
+        choice(f'tf_{i}', 4, label,
+               'Les abscisses ou instants demandés '
+               'doivent être définis dans les données correspondantes.',
+               YES_NO)
+    for i in range(1, 4):
+        add(f'fdeb{i}', 4, f'Seuil de débit du front {i} (m³/s)',
+            'Débit pour lequel à chaque abscisse, est enregistré le temps '
+            'de premier dépassement (fichier TRAJEC si demandé).',
+            positive=True)
+    choice('iovis', 4, 'Sorties complémentaires',
+           'Largeurs et sorties complémentaires hydrauliques et '
+           'sédimentaires.',
+           YES_NO)
+    return {field.key: field for field in result}
+
+
+FIELDS = fields()
+
+
+def canonical(field, value):
+    value = str(value).strip()
+    if field.choices and 'y' in field.choices:
+        aliases = {'o': 'y', 'y': 'y', 'n': 'n', 'm': 'm'}
+        return aliases.get(value.lower(), value)
+    return value
+
+
+def validate(field, value):
+    """Return a storage value or raise a user-facing validation error."""
+    value = canonical(field, value)
+    if field.kind == 'choice':
+        if value not in field.choices:
+            raise ValueError('Choisissez une valeur dans la liste.')
+        return value
+    if field.kind == 'options':
+        if len(value) != 3 or any(c not in '12' for c in value):
+            raise ValueError('Renseignez les trois choix de contrainte.')
+        return value
+    if field.kind == 'text':
+        return value
+    if not value and field.optional:
+        return value
+    value = value.replace(',', '.').replace('d', 'E').replace('D', 'E')
+    try:
+        if field.kind == 'time' and ':' in value:
+            parts = value.split(':')
+            valid = all(p.isascii() and p.isdigit() for p in parts)
+            if len(parts) != 4 or not valid:
+                raise ValueError(
+                    'Utilisez j:h:min:s ou un nombre de secondes.')
+            number = sum(Decimal(p) * factor for p, factor in
+                         zip(parts, (86400, 3600, 60, 1)))
+        else:
+            number = Decimal(value)
+    except InvalidOperation:
+        raise ValueError('Saisissez un nombre fini.') from None
+    if not number.is_finite():
+        raise ValueError('Saisissez un nombre fini.')
+    if number < field.minimum or (field.positive and number == 0):
+        raise ValueError('La valeur doit être strictement positive.'
+                         if field.positive else
+                         'La valeur doit être positive ou nulle.')
+    if field.maximum is not None and number > field.maximum:
+        raise ValueError(f'La valeur ne doit pas dépasser {field.maximum}.')
+    if field.key == 'rubarbe_sediment_por' and number >= 1:
+        raise ValueError('La porosité doit être inférieure à 1.')
+    scientific = not field.key.startswith('rubarbe_sediment_')
+    try:
+        Rubar3._real(str(number), field.key, scientific=scientific)
+    except (ValueError, DecimalException, OverflowError):
+        raise ValueError(
+            'Valeur trop longue ou hors plage pour le format du solveur '
+            '(12 colonnes).' if scientific else
+            'Valeur trop précise ou hors plage pour le format du solveur '
+            '(10 colonnes).') from None
+    return value
+
+
+def active(field, values):
+    """Dependencies only disable editing; stored values are never erased."""
+    key = field.key.removeprefix('rubarbe_')
+    transport = values.get('rubarbe_trased', 'n').lower() in ('y', 'o')
+    if key.startswith('sediment_') and not transport:
+        return False
+    if key == 'tiopdt':
+        return values.get('rubarbe_iopdt', '').lower() == 'm'
+    if key == 'sediment_tcadim':
+        return Rubar3.tcadim_required(
+            values.get('rubarbe_sediment_opts', ''),
+            values.get('rubarbe_sediment_option', ''))
+    if key in ('ros', 'dm', 'segma'):
+        return (not transport and
+                values.get('rubarbe_optfpc', '0') not in ('', '0'))
+    dependencies = {
+        'sediment_dcharg': ('sediment_odchar', ('0',)),
+        'sediment_halfa': ('sediment_odchar', ('1',)),
+        'sediment_dchard': ('sediment_demix', ('1',)),
+        'sediment_dchars': ('sediment_demix', ('1',)),
+        'sediment_varcons': ('sediment_defond', ('1',)),
+        'sediment_depot': ('sediment_defond', ('1',)),
+    }
+    if key in dependencies:
+        source, accepted = dependencies[key]
+        return values.get('rubarbe_' + source) in accepted
+    if key.startswith('fdeb'):
+        return values.get('rubarbe_tf_6', 'n').lower() in ('y', 'o')
+    return True

src/View/SolverParameters/RubarForm.py

@@ -0,0 +1,361 @@
+"""Grouped Rubar parameter editor, backed by the existing parameter list."""
+
+from functools import partial
+
+from PyQt5.QtCore import QCoreApplication, QEvent, Qt
+from PyQt5.QtGui import QKeySequence
+from PyQt5.QtWidgets import (
+    QComboBox, QFormLayout, QLabel, QLineEdit, QScrollArea,
+    QSizePolicy, QTabWidget, QVBoxLayout, QWidget,
+)
+
+from View.SolverParameters.RubarFields import (
+    FIELDS, GROUPS, OPTION_CHOICES, Field, active, canonical, validate,
+)
+from View.SolverParameters.UndoCommand import SetCommand
+
+
+def tr(text):
+    return QCoreApplication.translate('RubarParameters', text)
+
+
+class RubarForm(QWidget):
+    def __init__(self, parameters, undo, editable=True, parent=None):
+        super().__init__(parent)
+        self.parameters = parameters
+        self.stack = undo
+        self.editable = editable
+        self.editors = {}
+        self.labels = {}
+        self.errors = {}
+        self.descriptions = dict(FIELDS)
+        self.indices = {p.name: i for i, p in enumerate(parameters.parameters)}
+        self._refreshing = False
+        self._options_pending = False
+        layout = QVBoxLayout(self)
+        intro = QLabel(tr('Les valeurs valides sont appliquées à l’étude. '
+                          'Pensez à enregistrer l’étude pour les conserver.'))
+        intro.setWordWrap(True)
+        layout.addWidget(intro)
+        self.notice = QLabel()
+        self.notice.setWordWrap(True)
+        self.notice.setTextFormat(Qt.PlainText)
+        layout.addWidget(self.notice)
+        self.tabs = QTabWidget()
+        layout.addWidget(self.tabs)
+        forms = []
+        for group in GROUPS:
+            page = QWidget()
+            form = QFormLayout(page)
+            form.setFieldGrowthPolicy(QFormLayout.AllNonFixedFieldsGrow)
+            form.setRowWrapPolicy(QFormLayout.WrapLongRows)
+            form.setVerticalSpacing(12)
+            scroll = QScrollArea()
+            scroll.setWidgetResizable(True)
+            scroll.setWidget(page)
+            self.tabs.addTab(scroll, tr(group))
+            forms.append(form)
+        # Metadata order is stable even when stored parameter order changes.
+        keys = [key for key in FIELDS if key in self.indices]
+        keys += [key for key in self.indices if key not in FIELDS]
+        for key in keys:
+            field = self.descriptions.setdefault(key, Field(
+                key, 0, key, 'Paramètre supplémentaire conservé tel quel.',
+                kind='text'))
+            self._add_field(field, forms[field.group])
+        self.help = QLabel(tr(
+            'Sélectionnez un réglage pour afficher son aide.'))
+        self.help.setWordWrap(True)
+        self.help.setTextFormat(Qt.PlainText)
+        self.help.setMinimumHeight(60)
+        layout.addWidget(self.help)
+        self.stack.indexChanged.connect(self.refresh)
+        self.refresh()
+
+    def _add_field(self, field, form):
+        key = field.key
+        container = QWidget()
+        box = QVBoxLayout(container)
+        box.setContentsMargins(0, 0, 0, 0)
+        if field.kind == 'options':
+            editor = QWidget()
+            subform = QFormLayout(editor)
+            subform.setContentsMargins(0, 0, 0, 0)
+            self.option_editors = []
+            for title, choices, tooltip in OPTION_CHOICES:
+                combo = QComboBox()
+                combo.setSizeAdjustPolicy(
+                    QComboBox.AdjustToMinimumContentsLengthWithIcon)
+                combo.addItem(tr('À renseigner'), None)
+                for value, label in choices.items():
+                    combo.addItem(tr(label), value)
+                combo.currentIndexChanged.connect(self._commit_options)
+                combo.setProperty('parameter', key)
+                combo.installEventFilter(self)
+                combo.setToolTip(tr(tooltip) + '\n' + key)
+                subform.addRow(tr(title), combo)
+                self.option_editors.append(combo)
+        elif field.kind == 'choice':
+            editor = QComboBox()
+            editor.setSizeAdjustPolicy(
+                QComboBox.AdjustToMinimumContentsLengthWithIcon)
+            for value, label in field.choices.items():
+                editor.addItem(tr(label), value)
+            editor.currentIndexChanged.connect(
+                partial(self._choice_changed, key))
+        else:
+            editor = QLineEdit()
+            editor.setPlaceholderText(tr('Valeur du solveur') if field.optional
+                                      else tr('Saisir une valeur'))
+            editor.textEdited.connect(partial(self._check_draft, key))
+            editor.editingFinished.connect(partial(self._finish_text, key))
+        editor.setSizePolicy(QSizePolicy.Expanding, QSizePolicy.Preferred)
+        editor.setObjectName(key)
+        editor.setProperty('parameter', key)
+        editor.installEventFilter(self)
+        editor.setToolTip(
+            tr(field.help) + '\n' + key + '\nNotice §3.1 / §3.10')
+        box.addWidget(editor)
+        error = QLabel()
+        error.setWordWrap(True)
+        error.setTextFormat(Qt.PlainText)
+        error.setStyleSheet('color: #b42318;')
+        box.addWidget(error)
+        label = QLabel(tr(field.label))
+        label.setWordWrap(True)
+        label.setFixedWidth(310)
+        label.setMinimumHeight(label.heightForWidth(310))
+        label.setBuddy(editor)
+        label.setToolTip(editor.toolTip())
+        form.addRow(label, container)
+        self.editors[key] = editor
+        self.labels[key] = label
+        self.errors[key] = error
+
+    def eventFilter(self, watched, event):
+        if event.type() in (QEvent.FocusIn, QEvent.Enter):
+            key = watched.property('parameter')
+            if key:
+                field = self.descriptions[key]
+                self.help.setText(tr(field.help) + '\n' + key)
+        # Use the model undo stack even inside a line editor.
+        if event.type() in (QEvent.ShortcutOverride, QEvent.KeyPress):
+            if (event.matches(QKeySequence.Undo) or
+                    event.matches(QKeySequence.Redo)):
+                if event.type() == QEvent.KeyPress:
+                    if event.matches(QKeySequence.Undo):
+                        self.undo()
+                    else:
+                        self.redo()
+                event.accept()
+                return True
+        return super().eventFilter(watched, event)
+
+    def values(self):
+        return {p.name: str(p.value) for p in self.parameters.parameters}
+
+    def _error(self, key, message):
+        self.errors[key].setText(message)
+        self.errors[key].setVisible(bool(message))
+        self.editors[key].setStyleSheet(
+            'QLineEdit { border: 1px solid #b42318; }' if message else '')
+
+    def _validated(self, key, value):
+        try:
+            value = validate(self.descriptions[key], value)
+        except ValueError as error:
+            self._error(key, tr(str(error)) + tr(' Valeur non enregistrée.'))
+            return None
+        self._error(key, '')
+        return value
+
+    def _check_draft(self, key, value):
+        self._validated(key, value)
+
+    def _finish_text(self, key):
+        editor = self.editors[key]
+        if editor.isModified():
+            if self.commit(key, editor.text()):
+                editor.setModified(False)
+
+    def _choice_changed(self, key, index):
+        if not self._refreshing:
+            value = self.editors[key].itemData(index)
+            if value is not None:
+                self.commit(key, value)
+
+    def _commit_options(self):
+        if self._refreshing:
+            return
+        values = [combo.currentData() for combo in self.option_editors]
+        key = 'rubarbe_sediment_option'
+        if any(value is None for value in values):
+            self._options_pending = True
+            self._error(key, tr(
+                'Renseignez les trois choix. Valeur non enregistrée.'))
+        else:
+            self._options_pending = False
+            self.commit(key, ''.join(values))
+
+    def apply_pending(self):
+        """Commit focused text before Ctrl+S; leave invalid drafts visible."""
+        values = self.values()
+        if (values.get('rubarbe_condam') == '0' and
+                values.get('rubarbe_regime') == '1'):
+            self.tabs.setCurrentIndex(1)
+            self.editors['rubarbe_condam'].setFocus()
+            return False
+        for key, editor in self.editors.items():
+            if (isinstance(editor, QLineEdit) and editor.isModified() and
+                    active(self.descriptions[key], self.values())):
+                if not self.commit(key, editor.text()):
+                    self.tabs.setCurrentIndex(self.descriptions[key].group)
+                    editor.setFocus()
+                    return False
+        if (self._options_pending and
+                active(FIELDS['rubarbe_sediment_option'], self.values())):
+            self.tabs.setCurrentIndex(2)
+            self.option_editors[0].setFocus()
+            return False
+        return True
+
+    def commit(self, key, value):
+        field = self.descriptions[key]
+        if not self.editable or not active(field, self.values()):
+            return False
+        value = self._validated(key, value)
+        if value is None:
+            return False
+        values = self.values()
+        values[key] = value
+        if (key in ('rubarbe_condam', 'rubarbe_regime') and
+                values.get('rubarbe_condam') == '0' and
+                values.get('rubarbe_regime') == '1'):
+            self.refresh()
+            return False
+        if isinstance(self.editors[key], QLineEdit):
+            self.editors[key].setModified(False)
+        if self.values()[key] != value:
+            self.stack.push(SetCommand(self.parameters, self.indices[key],
+                                       'value', value))
+        elif isinstance(self.editors[key], QLineEdit):
+            self.editors[key].setText(value)
+        return True
+
+    def refresh(self, *args):
+        self._refreshing = True
+        values = self.values()
+        for key, editor in self.editors.items():
+            field = self.descriptions[key]
+            value = canonical(field, values[key])
+            enabled = active(field, values)
+            editor.setEnabled(self.editable and enabled)
+            self.labels[key].setEnabled(enabled)
+            if isinstance(editor, QLineEdit) and editor.isModified():
+                self._validated(key, editor.text())
+                continue
+            if field.kind == 'options' and self._options_pending:
+                self.errors[key].setVisible(enabled)
+                continue
+            if field.kind == 'options':
+                valid = len(value) == 3 and all(c in '12' for c in value)
+                for i, combo in enumerate(self.option_editors):
+                    index = combo.findData(value[i]) if valid else 0
+                    combo.setCurrentIndex(index)
+            elif isinstance(editor, QComboBox):
+                # Display unknown historical values without replacing them.
+                for i in reversed(range(editor.count())):
+                    if editor.itemData(i) is None:
+                        editor.removeItem(i)
+                index = editor.findData(value)
+                if index < 0:
+                    editor.addItem(tr('Valeur non reconnue : ') + value, None)
+                    index = editor.count() - 1
+                editor.setCurrentIndex(index)
+            else:
+                editor.setText(values[key])
+            try:
+                validate(field, values[key])
+                self._error(key, '')
+            except ValueError as error:
+                self._error(key, tr('Valeur enregistrée à vérifier : ') +
+                            values[key] + ' — ' + tr(str(error)))
+            self.errors[key].setVisible(
+                enabled and bool(self.errors[key].text()))
+        self._update_context(values)
+        for key, forbidden, blocked in (
+            ('rubarbe_regime', '1', values.get('rubarbe_condam') == '0'),
+            ('rubarbe_condam', '0', values.get('rubarbe_regime') == '1'),
+        ):
+            editor = self.editors.get(key)
+            if editor is None:
+                continue
+            index = editor.findData(forbidden)
+            if index >= 0:
+                editor.model().item(index).setEnabled(not blocked)
+        self._refreshing = False
+
+    def _update_context(self, values):
+        warnings = []
+        if not self.editable:
+            warnings.append(tr('Étude en lecture seule.'))
+        transport = values.get('rubarbe_trased', '').lower() in ('y', 'o')
+        if not transport:
+            warnings.append(tr('Transport solide désactivé : les réglages '
+                               'sédimentaires sont conservés mais inactifs.'))
+        elif (values.get('rubarbe_sediment_capsol') in ('2', '3') and
+              values.get('rubarbe_sediment_choixc') == '2'):
+            warnings.append(tr(
+                'La capacité solide choisie nécessite une méthode MPC ; '
+                'la répartition uniforme est incompatible.'))
+        if (values.get('rubarbe_condam') == '0' and
+                values.get('rubarbe_regime') == '1'):
+            warnings.append(tr('Une limite amont réfléchissante ne permet pas '
+                               'd’imposer aussi le tirant d’eau.'))
+        try:
+            from tools import old_pamhyr_date_to_timestamp
+
+            def seconds(value):
+                return (old_pamhyr_date_to_timestamp(value) if ':' in value
+                        else float(value.replace(',', '.')))
+
+            start = seconds(values['rubarbe_tinit'])
+            if seconds(values['rubarbe_tmax']) <= start:
+                warnings.append(tr('La fin du calcul doit suivre son début.'))
+        except (KeyError, ValueError):
+            pass
+        if transport:
+            try:
+                if float(values.get('rubarbe_sediment_visc', '0')) > 0.001:
+                    warnings.append(tr(
+                        'Viscosité élevée pour de l’eau : vérifiez la valeur. '
+                        'Le seuil critique adimensionnel se saisit dans son '
+                        'propre champ.'))
+            except ValueError:
+                pass
+            if (FIELDS['rubarbe_sediment_tcadim'].key in self.editors and
+                    active(FIELDS['rubarbe_sediment_tcadim'], values) and
+                    not values.get('rubarbe_sediment_tcadim', '').strip()):
+                warnings.append(tr(
+                    'La loi et les options choisies utilisent un seuil '
+                    'critique adimensionnel : renseignez-le.'))
+        self.notice.setText('\n'.join(warnings))
+        self.notice.setVisible(bool(warnings))
+
+    def undo(self):
+        if self.editable:
+            self._options_pending = False
+            for editor in self.editors.values():
+                if isinstance(editor, QLineEdit):
+                    editor.setModified(False)
+            self.stack.undo()
+            self.refresh()
+
+    def redo(self):
+        if self.editable:
+            self._options_pending = False
+            for editor in self.editors.values():
+                if isinstance(editor, QLineEdit):
+                    editor.setModified(False)
+            self.stack.redo()
+            self.refresh()

src/View/SolverParameters/Window.py

@@ -42,6 +42,7 @@ from PyQt5.QtWidgets import (
 
 from View.SolverParameters.UndoCommand import *
 from View.SolverParameters.Table import TableModel, ValueDelegate
+from View.SolverParameters.RubarForm import RubarForm
 from View.SolverParameters.translate import ParamTranslate
 
 from Solver.Solvers import solver_long_name, solver_type_list
@@ -62,7 +63,7 @@ class SolverParametersWindow(PamhyrWindow):
             title=name,
             study=study,
             config=config,
-            options=['copy'],
+            options=['undo', 'copy'],
             trad=trad,
             parent=parent
         )
@@ -88,6 +89,15 @@ class SolverParametersWindow(PamhyrWindow):
             editable_headers = []
 
         for st in solver_type_list:
+            if st in ('rubar3', 'rubarbe'):
+                widget = RubarForm(
+                    self._study.river.get_params(st), self._undo_stack[st],
+                    editable=self._study.is_editable(),
+                )
+                widget.setObjectName(f"tab_{st}")
+                self._table[st] = widget
+                self._tab_widget.addTab(widget, solver_long_name[st])
+                continue
             # Create widgets
             widget = QWidget()
             grid = QGridLayout()
@@ -135,8 +145,19 @@ class SolverParametersWindow(PamhyrWindow):
         tab = self.current_tab()
         self._table[tab].redo()
 
+    def _save(self, source):
+        for editor in self._table.values():
+            if isinstance(editor, RubarForm) and not editor.apply_pending():
+                self._tab_widget.setCurrentWidget(editor)
+                return
+        return super()._save(source)
+
     def _copy(self):
-        logger.info("TODO: copy")
+        editor = self.focusWidget()
+        if isinstance(editor, QLineEdit):
+            editor.copy()
 
     def _paste(self):
-        logger.info("TODO: paste")
+        editor = self.focusWidget()
+        if isinstance(editor, QLineEdit) and self._study.is_editable():
+            editor.paste()

src/View/ui/SolverParameters.ui

@@ -6,8 +6,8 @@
    <rect>
     <x>0</x>
     <y>0</y>
-    <width>900</width>
+    <width>1000</width>
-    <height>480</height>
+    <height>680</height>
    </rect>
   </property>
   <property name="windowTitle">

src/config.py

@@ -37,7 +37,7 @@ logger = logging.getLogger()
 
 class Config(SQL):
     def __init__(self):
-        self._version = '0.0.9'
+        self._version = '0.0.10'
         self.filename = Config.filename()
         self.set_default_value()
 
@@ -220,6 +220,26 @@ class Config(SQL):
                         "VALUES ('display_manning', 'False')"
                     )
 
+                if int(release) < 10:
+                    posix = os.name == 'posix'
+                    ext = "" if posix else ".exe"
+                    path = os.path.join("@install_dir",
+                                        "rubar", f"rubarbe{ext}")
+
+                    self.execute(f"""
+                      INSERT OR IGNORE INTO solver VALUES (
+                        'rubarbe',
+                        'default-RubarBE',
+                        'Default Pamhyr2 RubarBE version',
+
+                        '', '', '',
+
+                        '',
+                        '"{path}" @args @input',
+                        ''
+                      )
+                    """)
+
             self.execute(
                 f"UPDATE info SET value='{self._version}' " +
                 "WHERE key='version'"
@@ -416,11 +436,12 @@ class Config(SQL):
         new._cmd_solver = f""""{path}" @args @input"""
         self._solvers.append(new)
 
-        # Rubar3
+        # RubarBE (same executable as Rubar3, see notice §10.3)
-        ctor = solver_type_list["rubar3"]
+        ctor = solver_type_list["rubarbe"]
-        new = ctor("default-Rubar3")
+        new = ctor("default-RubarBE")
         new._description = "Default Pamhyr2 RubarBE version"
-        new._cmd_solver = f""""@install_dir/rubar/rubarbe{ext}" @args @input"""
+        path = os.path.join("@install_dir", "rubar", f"rubarbe{ext}")
+        new._cmd_solver = f""""{path}" @args @input"""
         self._solvers.append(new)
 
         # Mage fake 7