Fix PositionD parsing in config from strings like '123.1, 234.2', which had been doing the wrong thing.

.github/workflows/ci.yml

@@ -131,7 +131,6 @@ jobs:
                 brew link --overwrite fftw gcc wget || true
                 brew install eigen || true
                 brew link --overwrite eigen || true
-                brew install --cask gfortran || true
                 ls $FFTW_DIR
 
             - name: Download des_data
@@ -141,13 +140,6 @@ jobs:
                 if ! test -d $HOME/des_data || ! test -f $HOME/des_data/DECam_00154912_01.fits.fz; then wget --no-check-certificate http://www.sas.upenn.edu/~mjarvis/des_data.tar.gz; tar xfz des_data.tar.gz -C $HOME; fi
                 ln -s $HOME/des_data examples/des/
 
-            - name: Install numpy 2.0.0rc2
-              # Just on python 3.12, use the rc2 branch of numpy 2.0.0.
-              # XXX: Once the default numpy is >= 2.0, remove this.
-              if: matrix.py == 3.12
-              run: |
-                pip install numpy==2.0.0rc2
-
             - name: Install basic dependencies
               run: |
                 python -m pip install -U pip
@@ -167,16 +159,20 @@ jobs:
                 pip install -U matplotlib
 
             - name: Install astroplan
-              # astroplan isn't yet numpy 2.0 compatible.  So don't install that on 3.12.
-              if: matrix.py != 3.12
+              # astroplan isn't yet numpy 2.0 compatible.  So don't install that on 3.9+.
+              # I'm also getting errors with starlink, which I think may be related to
+              # numpy 2.0.
+              if: (matrix.py == 3.8) || (matrix.py == 3.9)
               run: |
                 pip install -U astroplan
+                pip install -U starlink-pyast
 
             - name: Install CPython-only dependencies
               if: matrix.py != 'pypy-3.7'
               run: |
                 # The only one left that doesn't seem to work right on pypy is starlink-pyast.
-                pip install -U starlink-pyast
+                # (This is now above in numpy 1.x versions.)
+                #pip install -U starlink-pyast
                 # And now pandas is acting up on pypy.
                 # cf. https://github.com/pandas-dev/pandas/issues/44253
                 pip install -U pandas

CHANGELOG.rst

@@ -127,3 +127,8 @@ Changes from v2.5.2 to v2.5.3
 - Added ScaleFlux and ScaleWavelength photon ops. (#1289)
 - Deprecated ChromaticObject.atRedshift. (#1291)
 - Various fixes to work with numpy 2.0. (#1297)
+
+Changes from v2.5.3 to v2.5.4
+=============================
+
+- Fixed a bug in the config layer parsing of PositionD from a string. (#1299)

galsim/config/value.py

@@ -451,8 +451,9 @@ def _GetPositionValue(param):
     """Convert a tuple or a string that looks like "a,b" into a galsim.PositionD.
     """
     try:
-        x = float(param[0])
-        y = float(param[1])
+        x, y = param
+        x = float(x)
+        y = float(y)
     except (ValueError, TypeError):
         try:
             x, y = param.split(',')

tests/test_config_value.py

@@ -1432,6 +1432,7 @@ def test_pos_value():
         'val1' : galsim.PositionD(0.1,0.2),
         'val2' : '0.1, 0.2',
         'val3' : None,
+        'val4' : '123.4, 567.8',
         'xy1' : { 'type' : 'XY', 'x' : 1.3, 'y' : 2.4 },
         'ran1' : { 'type' : 'RandomCircle', 'radius' : 3 },
         'ran2' : { 'type' : 'RandomCircle', 'radius' : 1, 'center' : galsim.PositionD(3,7) },
@@ -1487,6 +1488,10 @@ def test_pos_value():
     val3 = galsim.config.ParseValue(config,'val3',config, galsim.PositionD)[0]
     np.testing.assert_equal(val3, None)
 
+    val4 = galsim.config.ParseValue(config,'val4',config, galsim.PositionD)[0]
+    np.testing.assert_almost_equal(val4.x, 123.4)
+    np.testing.assert_almost_equal(val4.y, 567.8)
+
     xy1 = galsim.config.ParseValue(config,'xy1',config, galsim.PositionD)[0]
     np.testing.assert_almost_equal(xy1.x, 1.3)
     np.testing.assert_almost_equal(xy1.y, 2.4)
@@ -1594,6 +1599,7 @@ def test_pos_value():
     # Finally, these value got changed, so they won't match the original
     # unless we manually set them back to the original strings.
     clean_config['val2'] = '0.1, 0.2'
+    clean_config['val4'] = '123.4, 567.8'
     clean_config['cur2'] = '@input.val2'
     clean_config['input']['val2'] = '0.3, 0.4'
     assert clean_config == orig_config

tests/test_zernike.py

@@ -390,7 +390,7 @@ def Z11_grad(x, y):
     # fig.colorbar(scat2, ax=axes[2])
     # plt.show()
 
-    np.testing.assert_allclose(Z11.evalCartesianGrad(x, y), Z11_grad(x, y), rtol=0, atol=1e-12)
+    np.testing.assert_allclose(Z11.evalCartesianGrad(x, y), Z11_grad(x, y), rtol=1.e-12, atol=1e-12)
 
     Z28 = Zernike([0]*28+[1])
 
@@ -400,7 +400,7 @@ def Z28_grad(x, y):
         grady = -6*np.sqrt(14)*y*(5*x**4 - 10*x**2*y**2 + y**4)
         return gradx, grady
 
-    np.testing.assert_allclose(Z28.evalCartesianGrad(x, y), Z28_grad(x, y), rtol=0, atol=1e-12)
+    np.testing.assert_allclose(Z28.evalCartesianGrad(x, y), Z28_grad(x, y), rtol=1.e-12, atol=1e-12)
 
     # Now try some finite differences on a broader set of input