carray 3.0.1 → 3.0.3

data/ext/carray_attribute.c

@@ -13,6 +13,7 @@
 
 #include "carray.h"
 #include "ca_obj_face.h"   /* CA_OBJ_RECORD + rb_ca_record_get_data_class */
+#include "carray_internal.h"   /* rb_ca_inherit_attr */
 
 /* ------------------------------------------------------------------- */
 
@@ -436,6 +437,8 @@ rb_ca_is_object_type (VALUE self)
 /* ------------------------------------------------------------------- */
 
 static ID id_parent;
+static ID id_attr;
+static ID id_attrs;
 
 VALUE
 rb_ca_parent (VALUE self)
@@ -443,13 +446,41 @@ rb_ca_parent (VALUE self)
   return rb_ivar_get(self, id_parent);
 }
 
+/* Give dst the attributes (lib/carray/attribute.rb) that src shows, as its
+   own Hash.  Used by the methods that produce a new entity holding the
+   same values -- copy, to_type.  The parent walk here only decides
+   whether any array on the chain has an @attr at all; which of them
+   src actually shows (a lazy op view stops the chain) is left to #attrs,
+   so the rule lives in one place. */
+void
+rb_ca_inherit_attr (VALUE dst, VALUE src)
+{
+  VALUE obj = src, attrs;
+  int depth;
+
+  for (depth = 0; depth < 64 && ! NIL_P(obj); depth++) {
+    if ( rb_ivar_defined(obj, id_attr) ) {
+      break;
+    }
+    obj = rb_ivar_get(obj, id_parent);
+  }
+  if ( depth == 64 || NIL_P(obj) ) {
+    return;
+  }
+  attrs = rb_funcall(src, id_attrs, 0);
+  if ( RHASH_SIZE(attrs) > 0 ) {
+    rb_ivar_set(dst, id_attr, rb_hash_dup(attrs));
+  }
+}
+
+/* A view of a frozen array is not frozen itself: building it may still set
+   it up after this (an ivar, a lift that moves it onto another parent), and
+   writes through it are refused by ca_is_readonly, which walks to the
+   frozen root and finds CA_FLAG_READ_ONLY there. */
 VALUE
 rb_ca_set_parent (VALUE self, VALUE obj)
 {
   rb_ivar_set(self, id_parent, obj);
-  if ( OBJ_FROZEN(obj) ) {
-    rb_ca_freeze(self);
-  }
   return obj;
 }
 
@@ -521,6 +552,15 @@ ca_root_array (void *ap)
   }
 }
 
+/* Whether two arrays may share memory: their cells live in the same
+   root.  An operation that writes one while reading the other must read
+   from a copy taken before it writes. */
+int
+ca_shares_root (void *ap, void *aq)
+{
+  return ca_root_array(ap) == ca_root_array(aq);
+}
+
 static VALUE
 rb_ca_root_array (VALUE self)
 {
@@ -584,6 +624,8 @@ void
 Init_carray_attribute (void)
 {
   id_parent     = rb_intern("parent");
+  id_attr       = rb_intern("@attr");
+  id_attrs      = rb_intern("attrs");
 
   rb_define_method(rb_cCArray, "obj_type", rb_ca_obj_type, 0);
   rb_define_method(rb_cCArray, "data_type", rb_ca_data_type, 0);

data/ext/carray_bincount.c

@@ -72,6 +72,34 @@
   } \
 } while (0)
 
+/* Checked before a window opens: the dispatches above raise on a type
+   they do not cover, which would leave the labels (and weights) attached. */
+static void
+bincount_check_types (CArray *cl, CArray *cw)
+{
+  switch ( cl->data_type ) {
+  case CA_INT8:  case CA_INT16:  case CA_INT32:  case CA_INT64:
+  case CA_UINT8: case CA_UINT16: case CA_UINT32: case CA_UINT64:
+    break;
+  default:
+    rb_raise(rb_eCADataTypeError,
+             "bincount: integer label array required (got %d)",
+             cl->data_type);
+  }
+  if ( ! cw ) {
+    return;
+  }
+  switch ( cw->data_type ) {
+  case CA_FLOAT64: case CA_FLOAT32:
+  case CA_INT8:  case CA_INT16:  case CA_INT32:  case CA_INT64:
+  case CA_UINT8: case CA_UINT16: case CA_UINT32: case CA_UINT64:
+    break;
+  default:
+    rb_raise(rb_eCADataTypeError,
+             "bincount: weights must be numeric (got %d)", cw->data_type);
+  }
+}
+
 /* __bincount_count__(length) -- count occurrences of each label in
    self.  Allocates a zero-filled UInt32 (or UInt64 if length >= 2^32)
    output of size `length`, then runs the 8-way label dispatch. */
@@ -104,7 +132,8 @@ rb_ca_bincount_count_kernel (VALUE self, VALUE rlength)
     return vout;
   }
 
-  ca_attach(cl);
+  bincount_check_types(cl, NULL);
+  ca_attach(cl);   /* window: nothing raises inside */
   n      = cl->elements;
   mlabel = cl->mask ? (boolean8_t *) cl->mask->ptr : NULL;
 
@@ -213,7 +242,8 @@ rb_ca_bincount_weighted_kernel (VALUE self, VALUE rweights, VALUE rlength)
     return vout;
   }
 
-  ca_attach_n(2, cl, cw);
+  bincount_check_types(cl, cw);
+  ca_attach_n(2, cl, cw);   /* window: nothing raises inside */
   n       = cl->elements;
   mlabel  = cl->mask ? (boolean8_t *) cl->mask->ptr : NULL;
   mweight = cw->mask ? (boolean8_t *) cw->mask->ptr : NULL;

data/ext/carray_broadcast.c

@@ -23,13 +23,14 @@
         (right-to-left axis pairing; see the docstring at the function).
 
   Case B (cross-ndim dim-prepending) is not handled by the implicit
-  helpers; users with cross-ndim operands keep using the explicit :*
-  form or #broadcast_to (which does accept cross-ndim with target axes
-  pinned to size 1).  See PROPOSAL_BROADCASTING_AND_UNBOUND.md.
+  helpers; a cross-ndim operand has to declare the axis itself, either
+  with the newaxis sigil (`a + c[:_, nil]`) or with #broadcast_to (which
+  does accept cross-ndim with target axes pinned to size 1).
 
 ---------------------------------------------------------------------------- */
 
 #include "carray.h"
+#include "ca_obj_face.h"  /* CA_WRAPPER_LIFT */
 
 NORETURN(static void ca_broadcast_refuse_write (VALUE dst, VALUE src));
 
@@ -39,7 +40,7 @@ ca_broadcast_refuse_write (VALUE dst, VALUE src)
 {
   volatile VALUE dst_s = rb_inspect(rb_funcall(dst, rb_intern("shape"), 0));
   volatile VALUE src_s = rb_inspect(rb_funcall(src, rb_intern("shape"), 0));
-  rb_raise(rb_eRuntimeError,
+  rb_raise(rb_eArgError,
            "shape mismatch writing to carray (%s <- %s); shapes must agree "
            "once size-1 axes are dropped, or one side must be 1-D, or the "
            "source must be smaller and broadcastable -- use .flatten to "
@@ -119,7 +120,7 @@ ca_broadcast_view (VALUE src, int8_t ndim, ca_size_t *target_dim)
   TypedData_Get_Struct(src, CArray, &carray_data_type, cs);
 
   if (cs->ndim != ndim) {
-    rb_raise(rb_eRuntimeError,
+    rb_raise(rb_eArgError,
              "broadcast: ndim mismatch (%d vs %d)",
              (int) cs->ndim, (int) ndim);
   }
@@ -143,7 +144,7 @@ ca_broadcast_view (VALUE src, int8_t ndim, ca_size_t *target_dim)
       needs_broadcast = 1;
     }
     else {
-      rb_raise(rb_eRuntimeError,
+      rb_raise(rb_eArgError,
                "broadcast: cannot broadcast axis %d (%lld vs %lld)",
                i, (long long) cs->dim[i], (long long) target_dim[i]);
     }
@@ -172,20 +173,48 @@ ca_broadcast_view (VALUE src, int8_t ndim, ca_size_t *target_dim)
    raises). */
 NORETURN(static void ca_broadcast_refuse (VALUE self, VALUE other));
 
-/* Report a pair that cannot be brought to a common shape.  Naming both
-   shapes is the point of the message: the caller wrote two arrays that
-   cannot be combined, and the only remedy is to say which two. */
+/* Report a pair that cannot be brought to a common shape.  The message
+   names both shapes and points to the remedy that fits the pair: when
+   the ndim differs, the missing axes are added with :_; when the
+   element counts are equal, .flatten pairs the values in the order
+   they lie. */
 static void
 ca_broadcast_refuse (VALUE self, VALUE other)
 {
+  CArray *ca, *cb;
   volatile VALUE self_s  = rb_inspect(rb_funcall(self,  rb_intern("shape"), 0));
   volatile VALUE other_s = rb_inspect(rb_funcall(other, rb_intern("shape"), 0));
+  const char *flatten_hint;
+  int i;
+
+  TypedData_Get_Struct(self,  CArray, &carray_data_type, ca);
+  TypedData_Get_Struct(other, CArray, &carray_data_type, cb);
+
+  flatten_hint = (ca->elements == cb->elements)
+    ? "; to pair the values in the order they lie, use .flatten on both sides"
+    : "";
+
+  if (ca->ndim != cb->ndim) {
+    rb_raise(rb_eArgError,
+             "shape mismatch between operands (%s and %s): the operands "
+             "have different numbers of axes, and axes are not added "
+             "implicitly -- add size-1 axes with :_ so both have the same "
+             "number (x[:_, nil] makes a 1-D x a row, x[nil, :_] a column)%s",
+             StringValueCStr(self_s), StringValueCStr(other_s),
+             flatten_hint);
+  }
+
+  for (i = 0; i < ca->ndim; i++) {
+    if (ca->dim[i] != cb->dim[i] && ca->dim[i] != 1 && cb->dim[i] != 1) {
+      break;
+    }
+  }
   rb_raise(rb_eArgError,
-           "shape mismatch between operands (%s and %s); shapes must "
-           "agree, or differ only in size-1 axes with equal ndim -- use "
-           ".flatten on both sides to operate on the values in the order "
-           "they lie",
-           StringValueCStr(self_s), StringValueCStr(other_s));
+           "shape mismatch between operands (%s and %s): axis %d has "
+           "sizes %lld and %lld; each axis must have the same size, or "
+           "size 1 on one side%s",
+           StringValueCStr(self_s), StringValueCStr(other_s), i,
+           (long long) ca->dim[i], (long long) cb->dim[i], flatten_hint);
 }
 
 void
@@ -272,7 +301,7 @@ rb_ca_broadcast_to (int argc, VALUE *argv, VALUE self)
   target_ndim = (int8_t) argc;
 
   if (target_ndim < cs->ndim) {
-    rb_raise(rb_eRuntimeError,
+    rb_raise(rb_eArgError,
              "broadcast_to: target ndim %d smaller than source ndim %d",
              (int) target_ndim, (int) cs->ndim);
   }
@@ -303,6 +332,7 @@ rb_ca_broadcast_to (int argc, VALUE *argv, VALUE self)
     }
     obj = ca_wrap_struct(view);
     rb_ca_set_parent(obj, self);
+    CA_WRAPPER_LIFT(obj, self, cs);
     return obj;
   }
 
@@ -340,7 +370,7 @@ rb_ca_broadcast_to (int argc, VALUE *argv, VALUE self)
     else {
       /* sd == -1 (source exhausted) with dd > 1, or sd > 1 with
          dd > 1 and not equal: cannot broadcast. */
-      rb_raise(rb_eRuntimeError,
+      rb_raise(rb_eArgError,
                "broadcast_to: cannot broadcast axis %d "
                "(source %s, target %lld)",
                t_idx,
@@ -351,7 +381,7 @@ rb_ca_broadcast_to (int argc, VALUE *argv, VALUE self)
   }
 
   if (s_idx >= 0) {
-    rb_raise(rb_eRuntimeError,
+    rb_raise(rb_eArgError,
              "broadcast_to: %d source axes left unmatched", s_idx + 1);
   }
 
@@ -366,6 +396,8 @@ rb_ca_broadcast_to (int argc, VALUE *argv, VALUE self)
   }
   obj = ca_wrap_struct(view);
   rb_ca_set_parent(obj, self);
+  /* The Face stays on top, as for every view. */
+  CA_WRAPPER_LIFT(obj, self, cs);
   return obj;
 }