compatibility fixes for printf integer format specifiers Attempt to handle automatic conversion from integer floating point format in a way that is more compatible with Matlab behavior, including working properly for 64-bit integer values. * ov-base-diag.cc, ov-base-diag.h, ov-base-scalar.cc, ov-base-scalar.h, ov-base-sparse.cc, ov-base-sparse.h, ov-perm.cc, ov-perm.h, ov-range.cc, ov-range.h: Provide fast_elem_extract member function. * oct-stream.cc, oct-stream.h (printf_format_list::process_conversion): Ignore modifiers for integer formats. (printf_value_cache::curr_val): Store octave_value instead of NDArray. (printf_value_cache::data): Delete. (printf_value_cache::have_data): New member variable. (printf_value_cache::get_next_value): Rename from double_value, return individual value as an octave_value object instead of a double. (is_nan_or_inf, ok_for_signed_int_conv, ok_for_unsigned_int_conv, switch_to_g_format): New static functions. (DO_DOUBLE_CONV_1, DO_DOUBLE_CONV): Delete macros. (octave_base_stream::do_numeric_printf_conv): New function. (octave_base_stream::do_printf): Move code for handling numeric formats to do_numeric_printf_conv. * datestr.m: Round value for %d format.

--- a/libinterp/octave-value/ov-base-sparse.cc
+++ b/libinterp/octave-value/ov-base-sparse.cc
@@ -436,6 +436,20 @@
   return success;
 }
 
+
+template <class T>
+octave_value
+octave_base_sparse<T>::fast_elem_extract (octave_idx_type n) const
+{
+  octave_idx_type nr = matrix.rows ();
+  octave_idx_type nc = matrix.cols ();
+
+  octave_idx_type i = n % nr;
+  octave_idx_type j = n / nr;
+
+  return (i < nr && j < nc) ? octave_value (matrix(i,j)) : octave_value ();
+}
+
 template <class T>
 octave_value
 octave_base_sparse<T>::map (octave_base_value::unary_mapper_t umap) const