diff --git a/benchmark/benchmark.jl b/benchmark/benchmark.jl
index a8cafa8d..7f7fbc5c 100644
--- a/benchmark/benchmark.jl
+++ b/benchmark/benchmark.jl
@@ -21,7 +21,7 @@ function bench(;
     precompile = true,
     display = false,
     verbose = true,
-    discretization = :trapeze
+    discretization = :trapeze,
 )
 
     #######################################################
@@ -67,7 +67,7 @@ function bench(;
                 linear_solver = linear_solver,
                 max_iter = 0,
                 display = display,
-                discretization = discretization
+                discretization = discretization,
             )
             t_precomp += t
         end
@@ -86,7 +86,7 @@ function bench(;
             linear_solver = linear_solver,
             grid_size = grid_size,
             tol = tol,
-            discretization = discretization
+            discretization = discretization,
         )
         if !isnothing(problem[:obj]) && !isapprox(sol.objective, problem[:obj], rtol = 5e-2)
             error(
diff --git a/benchmark/prof.jl b/benchmark/prof.jl
index 68987f60..9d69d57c 100644
--- a/benchmark/prof.jl
+++ b/benchmark/prof.jl
@@ -39,7 +39,11 @@ if precompile
         CTDirect.DOCP_objective(CTDirect.DOCP_initial_guess(docp), docp)
     end
     if test_constraints
-        CTDirect.DOCP_constraints!(zeros(docp.dim_NLP_constraints), CTDirect.DOCP_initial_guess(docp), docp)
+        CTDirect.DOCP_constraints!(
+            zeros(docp.dim_NLP_constraints),
+            CTDirect.DOCP_initial_guess(docp),
+            docp,
+        )
     end
 end
 
@@ -49,8 +53,12 @@ if test_objective
     @btime CTDirect.DOCP_objective(CTDirect.DOCP_initial_guess(docp), docp)
 end
 if test_constraints
-    println("Timed constraints")   
-    @btime CTDirect.DOCP_constraints!(zeros(docp.dim_NLP_constraints), CTDirect.DOCP_initial_guess(docp), docp)
+    println("Timed constraints")
+    @btime CTDirect.DOCP_constraints!(
+        zeros(docp.dim_NLP_constraints),
+        CTDirect.DOCP_initial_guess(docp),
+        docp,
+    )
 end
 
 # transcription
@@ -62,10 +70,9 @@ end
 # full solve
 if test_solve
     println("Timed full solve")
-    @btime sol = direct_solve(ocp, grid_size = grid_size, display=false)
+    @btime sol = direct_solve(ocp, grid_size = grid_size, display = false)
 end
 
-
 if test_code_warntype
     if test_objective
         # NB. Pb with the mayer part: obj is type unstable (Any) because ocp.mayer is Union(Mayer,nothing), even for mayer problems (also, we should not even enter this code part for lagrange problems since has_mayer us defined as const in DOCP oO ...).
diff --git a/src/gauss.jl b/src/gauss.jl
index 6ddb02d4..cbc891c1 100644
--- a/src/gauss.jl
+++ b/src/gauss.jl
@@ -18,5 +18,11 @@ struct GaussLegendre2 <: Discretization
     butcher_a::Matrix{Float64}
     butcher_b::Vector{Float64}
     butcher_c::Vector{Float64}
-    GaussLegendre2() = new(2, 0, [0.25 (0.25 - sqrt(3)/6); (0.25 + sqrt(3)/6) 0.25], [0.5, 0.5], [(0.5 - sqrt(3)/6), (0.5 + sqrt(3)/6)])
+    GaussLegendre2() = new(
+        2,
+        0,
+        [0.25 (0.25-sqrt(3) / 6); (0.25+sqrt(3) / 6) 0.25],
+        [0.5, 0.5],
+        [(0.5 - sqrt(3) / 6), (0.5 + sqrt(3) / 6)],
+    )
 end
diff --git a/src/midpoint.jl b/src/midpoint.jl
index c5b56187..0ce639d5 100644
--- a/src/midpoint.jl
+++ b/src/midpoint.jl
@@ -12,19 +12,17 @@ struct Midpoint <: Discretization
     Midpoint() = new(1, 0)
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Retrieve state and control variables at given time step from the NLP variables.
 """
 function get_variables_at_time_step(xu, docp::DOCP{Midpoint}, i)
-
     nx = docp.dim_NLP_x
     n = docp.dim_OCP_x
     m = docp.dim_NLP_u
     N = docp.dim_NLP_steps
-    offset = (nx*(1+docp.discretization.stage) + m) * i
+    offset = (nx * (1 + docp.discretization.stage) + m) * i
 
     # retrieve scalar/vector OCP state (w/o lagrange state) 
     if n == 1
@@ -42,7 +40,7 @@ function get_variables_at_time_step(xu, docp::DOCP{Midpoint}, i)
     if i < N
         offset_u = offset
     else
-        offset_u = (nx*2 + m) * (i-1)
+        offset_u = (nx * 2 + m) * (i - 1)
     end
     if m == 1
         ui = xu[offset_u + nx + 1]
@@ -52,7 +50,7 @@ function get_variables_at_time_step(xu, docp::DOCP{Midpoint}, i)
 
     # retrieve vector stage variable (except at final time)
     if i < N
-        ki = xu[(offset + nx + m + 1):(offset + nx + m + nx) ]
+        ki = xu[(offset + nx + m + 1):(offset + nx + m + nx)]
     else
         ki = nothing
     end
@@ -60,17 +58,15 @@ function get_variables_at_time_step(xu, docp::DOCP{Midpoint}, i)
     return xi, ui, xli, ki
 end
 
-
 # internal NLP version for solution parsing
 # could be fused with one above if 
 # - using extended dynamics that include lagrange cost
 # - scalar case is handled at OCP level
 function get_NLP_variables_at_time_step(xu, docp, i, disc::Midpoint)
-
     nx = docp.dim_NLP_x
     m = docp.dim_NLP_u
     N = docp.dim_NLP_steps
-    offset = (nx*2 + m) * i
+    offset = (nx * 2 + m) * i
 
     # state
     xi = xu[(offset + 1):(offset + nx)]
@@ -78,12 +74,12 @@ function get_NLP_variables_at_time_step(xu, docp, i, disc::Midpoint)
     if i < N
         offset_u = offset
     else
-        offset_u = (nx*2 + m) * (i-1)
-    end 
+        offset_u = (nx * 2 + m) * (i - 1)
+    end
     ui = xu[(offset_u + nx + 1):(offset_u + nx + m)]
     # stage
     if i < N
-        ki = xu[(offset + nx + m + 1):(offset + nx + m + nx) ]
+        ki = xu[(offset + nx + m + 1):(offset + nx + m + nx)]
     else
         ki = nothing
     end
@@ -91,14 +87,12 @@ function get_NLP_variables_at_time_step(xu, docp, i, disc::Midpoint)
     return xi, ui, ki
 end
 
-
 function set_variables_at_time_step!(xu, x_init, u_init, docp, i, disc::Midpoint)
-
     nx = docp.dim_NLP_x
     n = docp.dim_OCP_x
     m = docp.dim_NLP_u
     N = docp.dim_NLP_steps
-    offset = (nx*2 + m) * i
+    offset = (nx * 2 + m) * i
 
     # NB. only set the actual state variables from the OCP 
     # - skip the possible additional state for lagrange cost
@@ -111,7 +105,6 @@ function set_variables_at_time_step!(xu, x_init, u_init, docp, i, disc::Midpoint
     end
 end
 
-
 # trivial version for now...
 # +++multiple dispatch here seems to cause more allocations !
 # +++? use abstract type for all Args ?
@@ -129,20 +122,19 @@ struct ArgsAtTimeStep_Midpoint
     next_time::Any
     next_state::Any
     next_lagrange_state::Any
-    
-    function ArgsAtTimeStep_Midpoint(xu, docp::DOCP{Midpoint}, v, time_grid, i::Int)
 
+    function ArgsAtTimeStep_Midpoint(xu, docp::DOCP{Midpoint}, v, time_grid, i::Int)
         disc = docp.discretization
 
         # variables
-        ti = time_grid[i+1]
+        ti = time_grid[i + 1]
         xi, ui, xli, ki = get_variables_at_time_step(xu, docp, i)
-        
+
         if i == docp.dim_NLP_steps
             return new(ti, xi, ui, xli, ki, disc)
         else
-            tip1 = time_grid[i+2]
-            xip1, uip1, xlip1 = get_variables_at_time_step(xu, docp, i+1)
+            tip1 = time_grid[i + 2]
+            xip1, uip1, xlip1 = get_variables_at_time_step(xu, docp, i + 1)
             return new(ti, xi, ui, xli, ki, tip1, xip1, xlip1)
         end
     end
@@ -151,20 +143,18 @@ function initArgs(xu, docp::DOCP{Midpoint}, time_grid)
     v = Float64[]
     docp.has_variable && (v = get_optim_variable(xu, docp))
     args = ArgsAtTimeStep_Midpoint(xu, docp, v, time_grid, 0)
-    return args, v 
+    return args, v
 end
 function updateArgs(args, xu, docp::DOCP{Midpoint}, v, time_grid, i::Int)
-    return ArgsAtTimeStep_Midpoint(xu, docp, v, time_grid, i+1)
+    return ArgsAtTimeStep_Midpoint(xu, docp, v, time_grid, i + 1)
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Set the constraints corresponding to the state equation
 """
 function setStateEquation!(docp::DOCP{Midpoint}, c, index::Int, args, v, i)
-
     ocp = docp.ocp
 
     # +++ later use butcher table in struct ?
@@ -181,36 +171,32 @@ function setStateEquation!(docp::DOCP{Midpoint}, c, index::Int, args, v, i)
     hi = tip1 - ti
 
     # midpoint rule
-    @. c[index:(index + docp.dim_OCP_x - 1)] =
-        xip1 - (xi + hi * ki[1:docp.dim_OCP_x])
+    @. c[index:(index + docp.dim_OCP_x - 1)] = xip1 - (xi + hi * ki[1:(docp.dim_OCP_x)])
     # +++ just define extended dynamics !
     if docp.has_lagrange
         c[index + docp.dim_OCP_x] = xlip1 - (xli + hi * ki[end])
     end
     index += docp.dim_NLP_x
-    
+
     # stage equation at mid-step
     t_s = 0.5 * (ti + tip1)
     x_s = 0.5 * (xi + xip1)
-    c[index:(index + docp.dim_OCP_x - 1)] .=
-        ki[1:docp.dim_OCP_x] .- ocp.dynamics(t_s, x_s, ui, v)
+    c[index:(index + docp.dim_OCP_x - 1)] .= ki[1:(docp.dim_OCP_x)] .- ocp.dynamics(t_s, x_s, ui, v)
     # +++ just define extended dynamics !
     if docp.has_lagrange
-        c[index + docp.dim_OCP_x] = ki[end] - ocp.lagrange(t_s, x_s, ui, v) 
+        c[index + docp.dim_OCP_x] = ki[end] - ocp.lagrange(t_s, x_s, ui, v)
     end
     index += docp.dim_NLP_x
 
     return index
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Set the path constraints at given time step
 """
 function setPathConstraints!(docp::DOCP{Midpoint}, c, index::Int, args, v, i::Int)
-
     ocp = docp.ocp
     ti = args.time
     xi = args.state
diff --git a/src/problem.jl b/src/problem.jl
index a55d26bb..825cc8f4 100644
--- a/src/problem.jl
+++ b/src/problem.jl
@@ -5,7 +5,6 @@
 # later we can add an option for control discretization: step or stage
 # (meaningful only for schemes with more than 1 stage, at least I will be able to compare the two ! further options may include CVP -control vector parametrization-, and maybe even pseudo-spectral ?)
 
-
 # generic discretization struct
 # NB. can we mutualize common fields at the abstract level ?
 abstract type Discretization end
@@ -70,7 +69,12 @@ struct DOCP{T <: Discretization}
     discretization::T
 
     # constructor
-    function DOCP(ocp::OptimalControlModel, grid_size::Integer, time_grid, discretization::Discretization)
+    function DOCP(
+        ocp::OptimalControlModel,
+        grid_size::Integer,
+        time_grid,
+        discretization::Discretization,
+    )
 
         # time grid
         if time_grid == nothing
@@ -120,7 +124,11 @@ struct DOCP{T <: Discretization}
         dim_stage = discretization.stage
 
         # NLP unknown (state + control + variable [+ stage])
-        dim_NLP_variables = (N + 1) * dim_NLP_x + (N + discretization.additional_controls) * dim_NLP_u + dim_NLP_v + N * dim_NLP_x * dim_stage
+        dim_NLP_variables =
+            (N + 1) * dim_NLP_x +
+            (N + discretization.additional_controls) * dim_NLP_u +
+            dim_NLP_v +
+            N * dim_NLP_x * dim_stage
 
         # NLP constraints 
         # parse NLP constraints (and initialize dimensions)
@@ -145,7 +153,10 @@ struct DOCP{T <: Discretization}
 
         # constraints (dynamics, stage, path, boundary, variable)
         dim_NLP_constraints =
-            N * (dim_NLP_x + (dim_NLP_x * dim_stage) + dim_path_cons) + dim_path_cons + dim_boundary_cons + dim_v_cons
+            N * (dim_NLP_x + (dim_NLP_x * dim_stage) + dim_path_cons) +
+            dim_path_cons +
+            dim_boundary_cons +
+            dim_v_cons
         if has_lagrange
             # add initial condition for lagrange state
             dim_NLP_constraints += 1
@@ -189,14 +200,13 @@ struct DOCP{T <: Discretization}
             Inf * ones(dim_NLP_variables),
             zeros(dim_NLP_constraints),
             zeros(dim_NLP_constraints),
-            discretization
+            discretization,
         )
 
         return docp
     end
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -207,14 +217,12 @@ function is_solvable(ocp)
     return solvable
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Build upper and lower bounds vectors for the DOCP nonlinear constraints.
 """
 function constraints_bounds!(docp::DOCP)
-
     lb = docp.con_l
     ub = docp.con_u
 
@@ -237,7 +245,6 @@ function constraints_bounds!(docp::DOCP)
     return lb, ub
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -270,14 +277,12 @@ function variables_bounds!(docp::DOCP)
     return var_l, var_u
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Compute the objective for the DOCP problem.
 """
 function DOCP_objective(xu, docp::DOCP)
-
     obj = 0.0
     N = docp.dim_NLP_steps
     ocp = docp.ocp
@@ -308,7 +313,6 @@ function DOCP_objective(xu, docp::DOCP)
     return obj
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -331,7 +335,6 @@ function DOCP_constraints!(c, xu, docp::DOCP)
         index = setPathConstraints!(docp, c, index, args, v, i)
         # update
         args = updateArgs(args, xu, docp, v, time_grid, i)
-
     end
 
     # path constraints at final time
@@ -345,14 +348,12 @@ function DOCP_constraints!(c, xu, docp::DOCP)
     return c
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Set bounds for the path constraints at given time step
 """
 function setPathBounds!(docp::DOCP, index::Int, lb, ub)
-
     ocp = docp.ocp
 
     # pure control constraints
@@ -379,14 +380,12 @@ function setPathBounds!(docp::DOCP, index::Int, lb, ub)
     return index
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Set the boundary and variable constraints
 """
 function setPointConstraints!(docp::DOCP, c, index::Int, xu, v)
-
     ocp = docp.ocp
 
     x0, u0, xl0 = get_variables_at_time_step(xu, docp, 0)
@@ -413,14 +412,12 @@ function setPointConstraints!(docp::DOCP, c, index::Int, xu, v)
     return index
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Set bounds for the boundary and variable constraints
 """
 function setPointBounds!(docp::DOCP, index::Int, lb, ub)
-
     ocp = docp.ocp
 
     # boundary constraints
@@ -447,7 +444,6 @@ function setPointBounds!(docp::DOCP, index::Int, lb, ub)
     return index
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -466,8 +462,15 @@ function DOCP_initial_guess(docp::DOCP, init::OptimalControlInit = OptimalContro
     # set state / control variables if provided
     time_grid = get_time_grid(NLP_X, docp)
     for i = 0:(docp.dim_NLP_steps)
-        ti = time_grid[i+1]
-        set_variables_at_time_step!(NLP_X, init.state_init(ti), init.control_init(ti), docp, i, docp.discretization)
+        ti = time_grid[i + 1]
+        set_variables_at_time_step!(
+            NLP_X,
+            init.state_init(ti),
+            init.control_init(ti),
+            docp,
+            i,
+            docp.discretization,
+        )
     end
 
     return NLP_X
diff --git a/src/solution.jl b/src/solution.jl
index 9bbd7ff7..46f0cbaf 100644
--- a/src/solution.jl
+++ b/src/solution.jl
@@ -188,7 +188,7 @@ function parse_DOCP_solution_dual(docp, multipliers, constraints)
         if i < N + 1
             P[i, :] = multipliers[i_m:(i_m + docp.dim_NLP_x - 1)]
             # skip dynamics constraints
-            i_c += docp.dim_NLP_x 
+            i_c += docp.dim_NLP_x
             i_m += docp.dim_NLP_x
             # skip stage constraints
             i_c += docp.dim_NLP_x * docp.discretization.stage
@@ -464,4 +464,3 @@ function set_box_block(T, mults, dim)
     end
     return t -> m_l(t), t -> m_u(t)
 end
-
diff --git a/src/solve.jl b/src/solve.jl
index 811d5c4e..8f7943a7 100644
--- a/src/solve.jl
+++ b/src/solve.jl
@@ -24,7 +24,7 @@ function direct_transcription(
     init = CTBase.__ocp_init(),
     grid_size = __grid_size(),
     time_grid = __time_grid(),
-    discretization = __discretization()
+    discretization = __discretization(),
 )
 
     # build DOCP
@@ -110,7 +110,7 @@ function direct_solve(
         init = init,
         grid_size = grid_size,
         time_grid = time_grid,
-        discretization = discretization
+        discretization = discretization,
     )
 
     # solve DOCP
@@ -136,4 +136,4 @@ weakdeps = Dict(IpoptBackend => :NLPModelsIpopt, MadNLPBackend => :MadNLP)
 
 function solve_docp(solver_backend::T, args...; kwargs...) where {T <: AbstractSolverBackend}
     throw(ExtensionError(weakdeps[T]))
-end
\ No newline at end of file
+end
diff --git a/src/trapeze.jl b/src/trapeze.jl
index e15db82e..18878300 100644
--- a/src/trapeze.jl
+++ b/src/trapeze.jl
@@ -3,7 +3,6 @@ Internal layout for NLP variables:
 [X_0,U_0, X_1,U_1, .., X_N,U_N, V]
 =#
 
-
 struct Trapeze <: Discretization
     stage::Int
     additional_controls::Int
@@ -11,14 +10,12 @@ struct Trapeze <: Discretization
     Trapeze() = new(0, 1)
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Retrieve state and control variables at given time step from the NLP variables. 
 """
 function get_variables_at_time_step(xu, docp::DOCP{Trapeze}, i)
-
     nx = docp.dim_NLP_x
     n = docp.dim_OCP_x
     m = docp.dim_NLP_u
@@ -46,13 +43,11 @@ function get_variables_at_time_step(xu, docp::DOCP{Trapeze}, i)
     return xi, ui, xli
 end
 
-
 # internal NLP version for solution parsing
 # could be fused with one above if 
 # - using extended dynamics that include lagrange cost
 # - scalar case is handled at OCP level
 function get_NLP_variables_at_time_step(xu, docp, i, disc::Trapeze)
-
     nx = docp.dim_NLP_x
     m = docp.dim_NLP_u
     offset = (nx + m) * i
@@ -65,9 +60,7 @@ function get_NLP_variables_at_time_step(xu, docp, i, disc::Trapeze)
     return xi, ui
 end
 
-
 function set_variables_at_time_step!(xu, x_init, u_init, docp, i, disc::Trapeze)
-
     nx = docp.dim_NLP_x
     n = docp.dim_OCP_x
     m = docp.dim_NLP_u
@@ -84,7 +77,6 @@ function set_variables_at_time_step!(xu, x_init, u_init, docp, i, disc::Trapeze)
     end
 end
 
-
 # ? use abstract type for args ?
 """
 $(TYPEDSIGNATURES)
@@ -102,7 +94,7 @@ struct ArgsAtTimeStep_Trapeze
     function ArgsAtTimeStep_Trapeze(xu, docp::DOCP{Trapeze}, v, time_grid, i::Int)
 
         # variables
-        ti = time_grid[i+1]
+        ti = time_grid[i + 1]
         xi, ui, xli = get_variables_at_time_step(xu, docp, i)
 
         # dynamics and lagrange cost
@@ -131,13 +123,12 @@ function updateArgs(args, xu, docp::DOCP{Trapeze}, v, time_grid, i)
     args_i, args_ip1 = args
     if i < docp.dim_NLP_steps - 1
         # are we allocating more than one args here ?
-        return (args_ip1, ArgsAtTimeStep_Trapeze(xu, docp, v, time_grid, i+2))
+        return (args_ip1, ArgsAtTimeStep_Trapeze(xu, docp, v, time_grid, i + 2))
     else
         return (args_ip1, args_ip1)
     end
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -164,7 +155,6 @@ function setStateEquation!(docp::DOCP{Trapeze}, c, index::Int, args, v, i)
     return index
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
diff --git a/src/utils.jl b/src/utils.jl
index e2844322..f4095b7d 100644
--- a/src/utils.jl
+++ b/src/utils.jl
@@ -5,14 +5,12 @@ Internal layout for NLP variables:
 with the convention u([t_i,t_i+1[) = U_i and u(tf) = U_N-1
 =#
 
-
 """
 $(TYPEDSIGNATURES)
 
 Retrieve optimization variables from the NLP variables.
 """
 function get_optim_variable(xu, docp)
-
     if docp.has_variable
         if docp.dim_NLP_v == 1
             return xu[end]
@@ -24,7 +22,6 @@ function get_optim_variable(xu, docp)
     end
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -38,7 +35,6 @@ function get_initial_time(xu, docp)
     end
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -52,7 +48,6 @@ function get_final_time(xu, docp)
     end
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -64,7 +59,6 @@ function get_time_grid(xu, docp)
     return @. t0 + docp.NLP_normalized_time_grid * (tf - t0)
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
@@ -74,14 +68,12 @@ function set_optim_variable!(xu, v_init, docp)
     xu[(end - docp.dim_NLP_v + 1):end] .= v_init
 end
 
-
 """
 $(TYPEDSIGNATURES)
 
 Build full, ordered sets of bounds for state, control or optimization variables
 """
 function build_bounds(dim_var, dim_box, box_triplet)
-
     x_lb = -Inf * ones(dim_var)
     x_ub = Inf * ones(dim_var)
     for j = 1:(dim_box)
@@ -93,7 +85,6 @@ function build_bounds(dim_var, dim_box, box_triplet)
     return x_lb, x_ub
 end
 
-
 # placeholders (see CTDirectExt)
 function export_ocp_solution end
 function import_ocp_solution end
diff --git a/test/suite/test_discretization.jl b/test/suite/test_discretization.jl
index 24961a2d..831a4c8a 100644
--- a/test/suite/test_discretization.jl
+++ b/test/suite/test_discretization.jl
@@ -1,6 +1,5 @@
 println("Test: discretization options")
 
-
 normalize_grid(t) = return (t .- t[1]) ./ (t[end] - t[1])
 
 # 1. simple integrator min energy (dual control for test)
@@ -64,13 +63,12 @@ end
     ocp = simple_integrator().ocp
     sol_t = direct_solve(ocp, display = false)
     sol_m = direct_solve(ocp, display = false, discretization = "midpoint")
-    @test sol_m.objective ≈ sol_t.objective rtol = 1e-2 
+    @test sol_m.objective ≈ sol_t.objective rtol = 1e-2
 end
 
 @testset verbose = true showtiming = true ":implicit_midpoint" begin
     ocp = double_integrator_freet0tf().ocp
     sol_t = direct_solve(ocp, display = false)
     sol_m = direct_solve(ocp, display = false, discretization = :midpoint)
-    @test sol_m.objective ≈ sol_t.objective rtol = 1e-2 
+    @test sol_m.objective ≈ sol_t.objective rtol = 1e-2
 end
-
diff --git a/test/suite/test_initial_guess.jl b/test/suite/test_initial_guess.jl
index 951a28fa..3884a1d8 100644
--- a/test/suite/test_initial_guess.jl
+++ b/test/suite/test_initial_guess.jl
@@ -29,22 +29,22 @@ u_vec = [0, 0.3, 0.1]
 @testset verbose = true showtiming = true ":default_init_no_arg" begin
     sol = direct_solve(ocp, display = false, max_iter = maxiter)
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->[0.1,0.1]).(T), rtol = 1e-2)
-    @test isapprox(sol.control.(T), (t->0.1).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> [0.1, 0.1]).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> 0.1).(T), rtol = 1e-2)
     @test isapprox(sol.variable, 0.1, rtol = 1e-2)
 end
 @testset verbose = true showtiming = true ":default_init_()" begin
     sol = direct_solve(ocp, display = false, init = (), max_iter = maxiter)
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->[0.1,0.1]).(T), rtol = 1e-2)
-    @test isapprox(sol.control.(T), (t->0.1).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> [0.1, 0.1]).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> 0.1).(T), rtol = 1e-2)
     @test isapprox(sol.variable, 0.1, rtol = 1e-2)
 end
 @testset verbose = true showtiming = true ":default_init_nothing" begin
     sol = direct_solve(ocp, display = false, init = nothing, max_iter = maxiter)
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->[0.1,0.1]).(T), rtol = 1e-2)
-    @test isapprox(sol.control.(T), (t->0.1).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> [0.1, 0.1]).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> 0.1).(T), rtol = 1e-2)
     @test sol.variable == 0.1
 end
 
@@ -52,12 +52,12 @@ end
 @testset verbose = true showtiming = true ":constant_x" begin
     sol = direct_solve(ocp, display = false, init = (state = x_const,), max_iter = maxiter)
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->x_const).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> x_const).(T), rtol = 1e-2)
 end
 @testset verbose = true showtiming = true ":constant_u" begin
     sol = direct_solve(ocp, display = false, init = (control = u_const,), max_iter = maxiter)
     T = sol.time_grid
-    @test isapprox(sol.control.(T), (t->u_const).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> u_const).(T), rtol = 1e-2)
 end
 @testset verbose = true showtiming = true ":constant_v" begin
     sol = direct_solve(ocp, display = false, init = (variable = v_const,), max_iter = maxiter)
@@ -71,8 +71,8 @@ end
         max_iter = maxiter,
     )
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->x_const).(T), rtol = 1e-2)
-    @test isapprox(sol.control.(T), (t->u_const).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> x_const).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> u_const).(T), rtol = 1e-2)
 end
 @testset verbose = true showtiming = true ":constant_xv" begin
     sol = direct_solve(
@@ -82,7 +82,7 @@ end
         max_iter = maxiter,
     )
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->x_const).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> x_const).(T), rtol = 1e-2)
     @test sol.variable == v_const
 end
 @testset verbose = true showtiming = true ":constant_uv" begin
@@ -93,7 +93,7 @@ end
         max_iter = maxiter,
     )
     T = sol.time_grid
-    @test isapprox(sol.control.(T), (t->u_const).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> u_const).(T), rtol = 1e-2)
     @test sol.variable == v_const
 end
 @testset verbose = true showtiming = true ":constant_xuv" begin
@@ -104,8 +104,8 @@ end
         max_iter = maxiter,
     )
     T = sol.time_grid
-    @test isapprox(sol.state.(T), (t->x_const).(T), rtol = 1e-2)
-    @test isapprox(sol.control.(T), (t->u_const).(T), rtol = 1e-2)
+    @test isapprox(sol.state.(T), (t -> x_const).(T), rtol = 1e-2)
+    @test isapprox(sol.control.(T), (t -> u_const).(T), rtol = 1e-2)
     @test sol.variable == v_const
 end
 
@@ -164,7 +164,7 @@ end
         init = (time = t_vec, state = x_matrix, control = u_vec, variable = v_const),
         max_iter = maxiter,
     )
-    @test isapprox(stack(sol.state.(t_matrix),dims=1), x_matrix, rtol = 1e-2)
+    @test isapprox(stack(sol.state.(t_matrix), dims = 1), x_matrix, rtol = 1e-2)
     @test isapprox(sol.control.(t_vec), u_vec, rtol = 1e-2)
     @test sol.variable == v_const
 end