Finish moving vectorized plot routines to pymathics.vectorizedplot

.github/workflows/plot-tests.yml

@@ -38,6 +38,12 @@ jobs:
         pip install -e .
         bash -x ./admin-tools/make-JSON-tables.sh
         remake -x develop-full
+    - name: install Mathic3 vectorizedplot Module
+      run: |
+        git clone --depth 1 https://github.com/Mathics3/Mathics3-Module-vectorizedplot.git
+        cd Mathics3-Module-vectorizedplot
+        python -m pip install --no-build-isolation -e .
+        cd ..
     - name: Test Mathics
       run: |
         make plot-detailed-tests

mathics/builtin/drawing/plot.py

@@ -11,7 +11,6 @@
 import palettable
 
 import mathics.eval.drawing.plot3d
-import mathics.eval.drawing.plot3d_vectorized
 from mathics.builtin.graphics import Graphics
 from mathics.builtin.options import options_to_rules
 from mathics.core.atoms import Integer, Integer0, MachineReal, String
@@ -39,29 +38,11 @@
 from mathics.eval.drawing.plot import get_plot_range
 from mathics.eval.nevaluator import eval_N
 
-# The vectorized plot function generates GraphicsComplex using NumericArray,
-# which no consumer will currently understand. So lets make it opt-in for now.
-# If it remains opt-in we'll probably want some combination of env variables,
-# Set option such as $UseVectorizedPlot, and maybe a non-standard Plot3D option.
-# For now an env variable is simplest.
-# TODO: work out exactly how to deploy.
 
-
-# can be set via environment variable at startup time,
-# or changed dynamically by setting the use_vectorized_plot flag
-use_vectorized_plot = os.getenv("MATHICS3_USE_VECTORIZED_PLOT", False)
-
-
-# get the plot eval function for the given class,
-# depending on whether vectorized plot functions are enabled
+# get the plot eval function for the given class
 def get_plot_eval_function(cls):
     function_name = "eval_" + cls.__name__
-    plot_module = (
-        mathics.eval.drawing.plot3d_vectorized
-        if use_vectorized_plot
-        else mathics.eval.drawing.plot3d
-    )
-    fun = getattr(plot_module, function_name)
+    fun = getattr(mathics.eval.drawing.plot3d, function_name)
     return fun
 
 
@@ -519,17 +500,9 @@ def error(*args, **kwargs):
             evaluation.message(builtin.get_name(), "invmaxrec", max_depth, 15)
         self.max_depth = max_depth
 
-        # PlotRange option
-        # Normalize to always return a list of either numeric,
-        # Symbol (for vectorized plots), or str (for classic plogs)
-        # TODO: convert classic plots to use Symbol instead of str also
-        if use_vectorized_plot:
-            preserve_symbols = True
-            symbol_type = Symbol
-        else:
-            # TODO: get rid of these
-            preserve_symbols = False
-            symbol_type = str
+        # TODO: get rid of these
+        preserve_symbols = False
+        symbol_type = str
         plot_range = builtin.get_option(options, str(SymbolPlotRange), evaluation)
         plot_range = eval_N(plot_range, evaluation).to_python(
             preserve_symbols=preserve_symbols

mathics/builtin/drawing/plot_plot.py

@@ -18,7 +18,6 @@
 from mathics.core.symbols import SymbolTrue
 from mathics.core.systemsymbols import SymbolLogPlot, SymbolPlotRange, SymbolSequence
 from mathics.eval.drawing.plot import eval_Plot
-from mathics.eval.drawing.plot_vectorized import eval_Plot_vectorized
 
 from . import plot
 
@@ -78,20 +77,18 @@ def eval(self, functions, ranges, evaluation: Evaluation, options: dict):
         except ValueError:
             return None
 
-        # for classic plot we cache results, but for vectorized we can't
-        # because ndarray is unhashable, and in any case probably isn't useful
+        # cache results
         # TODO: does caching results in the classic case have demonstrable performance benefit?
         apply_function = self.apply_function
-        if not plot.use_vectorized_plot:
-            apply_function = lru_cache(apply_function)
+        apply_function = lru_cache(apply_function)
 
         # additional options specific to this class
         plot_options.functions = self.get_functions_param(functions)
         plot_options.apply_function = apply_function
         plot_options.use_log_scale = self.use_log_scale
         plot_options.expect_list = self.expect_list
         if plot_options.plot_points is None:
-            default_plot_points = 1000 if plot.use_vectorized_plot else 57
+            default_plot_points = 57
             plot_options.plot_points = default_plot_points
 
         # pass through the expanded plot_range options
@@ -102,8 +99,7 @@ def eval(self, functions, ranges, evaluation: Evaluation, options: dict):
         if plot_options.use_log_scale:
             options[str(SymbolLogPlot)] = SymbolTrue
 
-        # this will be either the vectorized or the classic eval function
-        eval_function = eval_Plot_vectorized if plot.use_vectorized_plot else eval_Plot
+        eval_function = eval_Plot
         with np.errstate(all="ignore"):  # suppress numpy warnings
             graphics = eval_function(plot_options, options, evaluation)
         return graphics

mathics/builtin/drawing/plot_plot3d.py

@@ -111,7 +111,7 @@ def eval(
 
         # supply default value
         if plot_options.plot_points is None:
-            default_plot_points = (200, 200) if plot.use_vectorized_plot else (7, 7)
+            default_plot_points = (7, 7)
             plot_options.plot_points = default_plot_points
 
         # subclass must set eval_function and graphics_class