1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
|
# Copyright (C) 2003 Vladimir Prus
# Use, modification, and distribution is subject to the Boost Software
# License, Version 1.0. (See accompanying file LICENSE_1_0.txt or copy
# at http://www.boost.org/LICENSE_1_0.txt)
# This module defines a class which allows to order arbitrary object with
# regard to arbitrary binary relation.
#
# The primary use case is the gcc toolset, which is sensitive to library order:
# if library 'a' uses symbols from library 'b', then 'a' must be present before
# 'b' on the linker's command line.
#
# This requirement can be lifted for gcc with GNU ld, but for gcc with Solaris
# LD (and for Solaris toolset as well), the order always matters.
#
# So, we need to store order requirements and then order libraries according to
# them. It is not possible to use the dependency graph as order requirements.
# What we need is a "use symbols" relationship while dependency graph provides
# the "needs to be updated" relationship.
#
# For example::
# lib a : a.cpp b;
# lib b ;
#
# For static linking, library 'a' need not depend on 'b'. However, it should
# still come before 'b' on the command line.
class order
{
rule __init__ ( )
{
}
# Adds the constraint that 'first' should preceede 'second'.
rule add-pair ( first second )
{
.constraits += $(first)--$(second) ;
}
NATIVE_RULE class@order : add-pair ;
# Given a list of objects, reorder them so that the constraints specified by
# 'add-pair' are satisfied.
#
# The algorithm was adopted from an awk script by Nikita Youshchenko
# (yoush at cs dot msu dot su)
rule order ( objects * )
{
# The algorithm used is the same is standard transitive closure, except
# that we're not keeping in-degree for all vertices, but rather removing
# edges.
local result ;
if $(objects)
{
local constraints = [ eliminate-unused-constraits $(objects) ] ;
# Find some library that nobody depends upon and add it to the
# 'result' array.
local obj ;
while $(objects)
{
local new_objects ;
while $(objects)
{
obj = $(objects[1]) ;
if [ has-no-dependents $(obj) : $(constraints) ]
{
# Emulate break ;
new_objects += $(objects[2-]) ;
objects = ;
}
else
{
new_objects += $(obj) ;
obj = ;
objects = $(objects[2-]) ;
}
}
if ! $(obj)
{
errors.error "Circular order dependencies" ;
}
# No problem with placing first.
result += $(obj) ;
# Remove all contraints where 'obj' comes first, since they are
# already satisfied.
constraints = [ remove-satisfied $(constraints) : $(obj) ] ;
# Add the remaining objects for further processing on the next
# iteration
objects = $(new_objects) ;
}
}
return $(result) ;
}
NATIVE_RULE class@order : order ;
# Eliminate constraints which mention objects not in 'objects'. In
# graph-theory terms, this is finding a subgraph induced by ordered
# vertices.
rule eliminate-unused-constraits ( objects * )
{
local result ;
for local c in $(.constraints)
{
local m = [ MATCH (.*)--(.*) : $(c) ] ;
if $(m[1]) in $(objects) && $(m[2]) in $(objects)
{
result += $(c) ;
}
}
return $(result) ;
}
# Returns true if there's no constraint in 'constaraints' where 'obj' comes
# second.
rule has-no-dependents ( obj : constraints * )
{
local failed ;
while $(constraints) && ! $(failed)
{
local c = $(constraints[1]) ;
local m = [ MATCH (.*)--(.*) : $(c) ] ;
if $(m[2]) = $(obj)
{
failed = true ;
}
constraints = $(constraints[2-]) ;
}
if ! $(failed)
{
return true ;
}
}
rule remove-satisfied ( constraints * : obj )
{
local result ;
for local c in $(constraints)
{
local m = [ MATCH (.*)--(.*) : $(c) ] ;
if $(m[1]) != $(obj)
{
result += $(c) ;
}
}
return $(result) ;
}
}
rule __test__ ( )
{
import "class" : new ;
import assert ;
c1 = [ new order ] ;
$(c1).add-pair l1 l2 ;
assert.result l1 l2 : $(c1).order l1 l2 ;
assert.result l1 l2 : $(c1).order l2 l1 ;
$(c1).add-pair l2 l3 ;
assert.result l1 l2 : $(c1).order l2 l1 ;
$(c1).add-pair x l2 ;
assert.result l1 l2 : $(c1).order l2 l1 ;
assert.result l1 l2 l3 : $(c1).order l2 l3 l1 ;
# The output should be stable for unconstrained
# elements.
assert.result l4 l5 : $(c1).order l4 l5 ;
}
|
