Fix blunder in operator priority

-a//-b did the wrong thing. Use -(a//-b) and add corresponding regression tests.

lslopt/lslbasefuncs.py

@@ -679,7 +679,7 @@ def div(a, b, f32=True):
                     return a # this could be handled by using S32 but it's probably faster this way
                 if (a < 0) ^ (b < 0):
                     # signs differ - Python rounds towards -inf, we need rounding towards 0
-                    return - a//-b # that's -(a//-b) not (-a)//-b
+                    return -(a//-b)
                 return a//b
             ret = F32(ff(a)/ff(b), f32)
             if math.isnan(ret): # A NaN result gives a math error.

testfuncs.py

@@ -516,6 +516,15 @@ def do_tests():
     shouldexcept('div(1, NaN)', ELSLMathError)
     shouldexcept('div(F32(1e40), F32(1e40))', ELSLMathError)
     shouldexcept('zstr("blah")', ELSLInvalidType)
+    test('div(1, 9)', 0)
+    test('div(8, 9)', 0)
+    test('div(9, 9)', 1)
+    test('div(-1,9)', 0)
+    test('div(-8,9)', 0)
+    test('div(-9,9)', -1)
+    test('div(1,-9)', 0)
+    test('div(8,-9)', 0)
+    test('div(9,-9)', -1)
     test(r'zstr(Key(u"xy\0zzy"))', Key(u'xy'))
     test('typecast(Infinity, unicode)', u'Infinity')
     test('typecast(NaN, unicode)', u'NaN')