Mercurial > octave
comparison liboctave/array/dim-vector.cc @ 21573:f3f8e1d3e399 stable
avoid mulitple definitions of static function-scope vars (bug #47372)
Some systems (Windows, others?) may generate multiple definitions of
function-scope static variables if those functions appear in header
files. This can cause trouble in various ways. Attempt to avoid the
problem by moving functions that contain file-scope varaibles from
header files to source files, or by using class-scope static variables
instead. Also attempt to avoid similar problems caused by implicit
instantiations of templates that have functions with static variables.
* Cell.h, Cell.cc (Cell::resize_fill_value):
Move definition from header to source file.
* oct-map.h, oct-map.cc (octave_fields::nil_rep): Likewise.
* ov-base-scalar.h, ov-base-scalar.cc (octave_base_scalar<ST>::dims):
Likewise.
* ov-fcn-handle.h, ov-fcn-handle.cc (octave_fcn_handle::dims): Likewise.
* ov.h, ov.cc (octave_value::nil_rep): Likewise.
* Array.h, Array.cc (Array<T>::nil_rep): Likewise.
* DiagArray2.h, DiagArray2.cc (DiagArray2<T>::elem,
DiagArray2<T>::checkelem): Likewise.
* Sparse.h, Sparse.cc (Sparse<T>::nil_rep): Likewise.
* dim-vector.h, dim-vector.cc (dim_vector::nil_rep): Likewise.
* idx-vector.h, idx-vector.cc (idx_vector::nil_rep,
idx_vector::err_rep): Likewise.
* oct-inttypes.h, oct-inttypes.cc (octave_int_base<T>::convert_real):
Likewise.
* symtab.h, symtab.cc (octave_value symbol_table::dummy_octave_value,
symbol_table::symbol_record symbol_table::dummy_symbol_record): New
class-scope static variables. Use them to eliminate function-scope
static variables.
* ov-int8.cc, ov-int16.cc, ov-int32.cc, ov-int64.cc, ov-uint8.cc,
ov-uint16.cc, ov-uint32.cc, ov-uint64.cc, ov-scalar.cc,
ov-flt-complex.cc, ov-complex.cc, ov-classdef.cc, ov-bool.cc,
Array-jit.cc, Array-os.cc, Array-tc.cc, Array-C.cc, Array-b.cc,
Array-ch.cc, Array-d.cc, Array-f.cc, Array-fC.cc, Array-i.cc,
Array-idx-vec.cc, Array-s.cc, Array-str.cc, Array-voidp.cc:
Prevent impliticit template instantiation.
* Array-tc.cc: Also instantiate Array<cdef_object>.
author | John W. Eaton <jwe@octave.org> |
---|---|
date | Fri, 01 Apr 2016 12:04:04 -0400 |
parents | 4197fc428c7d |
children | ae4d7dfea337 |
comparison
equal
deleted
inserted
replaced
21572:8fc45dd1cac6 | 21573:f3f8e1d3e399 |
---|---|
26 #endif | 26 #endif |
27 | 27 |
28 #include <iostream> | 28 #include <iostream> |
29 | 29 |
30 #include "dim-vector.h" | 30 #include "dim-vector.h" |
31 | |
32 octave_idx_type * | |
33 dim_vector::nil_rep (void) | |
34 { | |
35 static dim_vector zv (0, 0); | |
36 return zv.rep; | |
37 } | |
31 | 38 |
32 // The maximum allowed value for a dimension extent. This will normally be a | 39 // The maximum allowed value for a dimension extent. This will normally be a |
33 // tiny bit off the maximum value of octave_idx_type. | 40 // tiny bit off the maximum value of octave_idx_type. |
34 // Currently 1 is subtracted to allow safe conversion of any 2D Array into | 41 // Currently 1 is subtracted to allow safe conversion of any 2D Array into |
35 // Sparse, but this offset may change in the future. | 42 // Sparse, but this offset may change in the future. |