aboutsummaryrefslogtreecommitdiffhomepage
path: root/site/docs/cpp.md
blob: 3566dc9f3a9c13296e632c778cb05377b9e6f34f (plain)
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
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
---
layout: documentation
---

C++ Basics
==========

Use fully qualified include paths
---------------------------------

Includes are relative to the root of your workspace. For example, suppose
you have the following directory structure:

```
[workspace]/
    WORKSPACE
    a/
        BUILD
        a.h
        a.cc
    b/
        BUILD
        b.h
        b.cc
        main.cc
```

If `b/main.cc` needs to include b.h then we'd create the following `b/BUILD`
file:

```python
cc_library(
    name = "b",
    srcs = ["b.cc"],
    hdrs = ["b.h"],
)

cc_binary(
    name = "main",
    srcs = ["main.cc"],
    deps = [":b"],
)
```

`b/main.cc` would have the following include statement:

```cpp
#include "b/b.h"
```

Note that the full path from the package root is used. If we want `b/main.cc` to
also depend on `a/a.h`, we'd add the rule to `a/BUILD`:

```python
cc_library(
    name = "a",
    srcs = ["a.cc"],
    hdrs = ["a.h"],
    visibility = ["//b:__pkg__"],
)
```

Then we'd add a dependency to `b/BUILD`:

```python
cc_binary(
    name = "main",
    srcs = ["main.cc"],
    deps = [
        ":b",
        "//a",
    ],
)
```

And the following include to `b/main.cc`:

```cpp
#include "a/a.h"
```

`b/main.cc` will then be able to access symbols from `a/a.h` or `b/b.h`.

Transitive includes
-------------------

If a file includes a header then the file's rule should depend on that header's
library.  Conversely, only direct dependencies need to be specified as
dependencies.  For example, suppose `sandwich.h` includes `bread.h` and
`bread.h` includes `flour.h`.  `sandwich.h` doesn't include `flour.h` (who wants
flour in their sandwich?), so the BUILD file would look like:

```python
cc_library(
    name = "sandwich",
    srcs = ["sandwich.cc"],
    hdrs = ["sandwich.h"],
    deps = [":bread"],
)

cc_library(
    name = "bread",
    srcs = ["bread.cc"],
    hdrs = ["bread.h"],
    deps = [":flour"],
)

cc_library(
    name = "flour",
    srcs = ["flour.cc"],
    hdrs = ["flour.h"],
)
```

This expresses that the `sandwich` library depends on the `bread` library,
which depends on the `flour` library.

Adding include paths
--------------------

Sometimes you cannot (or do not want to) base include paths at the workspace
root. Existing libaries might already have a include directory that doesn't
match its path in your workspace.  For example, suppose you have the following
directory structure:

```
[workspace]/
    WORKSPACE
    third_party/
        some_lib/
            include/
                some_lib.h
            BUILD
            some_lib.cc
```

Bazel will expect `some_lib.h` to be included as
`third_party/some_lib/include/some_lib.h`, but suppose `some_lib.cc` includes
`"include/some_lib.h"`.  To make that include path valid,
`third_party/some_lib/BUILD` will need to specify that the `some_lib/`
directory is an include directory:

```python
cc_library(
    name = "some_lib",
    srcs = ["some_lib.cc"],
    hdrs = ["some_lib.h"],
    copts = ["-Ithird_party/some_lib"],
)
```

This is especially useful for external dependencies, as their header files
must otherwise be included with an `external/[repository-name]/` prefix.

Including external libraries: an example
----------------------------------------

Suppose you are using [Google Test](https://code.google.com/p/googletest/). You
can use one of the `new_` repository functions in the `WORKSPACE` file to
download Google Test and make it available in your repository:

```python
new_http_archive(
    name = "gtest",
    url = "https://googletest.googlecode.com/files/gtest-1.7.0.zip",
    sha256 = "247ca18dd83f53deb1328be17e4b1be31514cedfc1e3424f672bf11fd7e0d60d",
    build_file = "gtest.BUILD",
)
```

Then create `gtest.BUILD`, a BUILD file to use to compile Google Test.
Google Test has several "special" requirements that make its `cc_library` rule
more complicated:

* `gtest-1.7.0/src/gtest-all.cc` `#include`s all of the other files in
  `gtest-1.7.0/src/`, so we need to exclude it from the compile or we'll get
  link errors for duplicate symbols.
* It uses header files that relative to the `gtest-1.7.0/include/` directory
  (`"gtest/gtest.h"`), so we must add that directory the include paths.
* It uses "private" header files in `src/`, so we add that to the include pahs,
  too, so it can `#include "src/gtest-internal-inl.h"`.
* It needs to link in pthread, so we add that as a `linkopt`.

The final rule looks like this:

```python
cc_library(
    name = "main",
    srcs = glob(
        ["gtest-1.7.0/src/*.cc"],
        exclude = ["gtest-1.7.0/src/gtest-all.cc"]
    ),
    hdrs = glob(["gtest-1.7.0/include/**/*.h"]),
    copts = [
        "-Iexternal/gtest/gtest-1.7.0",
        "-Iexternal/gtest/gtest-1.7.0/include"
    ],
    linkopts = ["-pthread"],
    visibility = ["//visibility:public"],
)
```

This is somewhat messy: everything is prefixed with gtest-1.7.0 as a byproduct
of the archive's structure. You can make `new_http_archive` strip this prefix by
adding the `strip_prefix` attribute:

```python
new_http_archive(
    name = "gtest",
    url = "https://googletest.googlecode.com/files/gtest-1.7.0.zip",
    sha256 = "247ca18dd83f53deb1328be17e4b1be31514cedfc1e3424f672bf11fd7e0d60d",
    build_file = "gtest.BUILD",
    strip_prefix = "gtest-1.7.0",
)
```

Then `gtest.BUILD` would look like this:

```python
cc_library(
    name = "main",
    srcs = glob(
        ["src/*.cc"],
        exclude = ["src/gtest-all.cc"]
    ),
    hdrs = glob(["include/**/*.h"]),
    copts = [
        "-Iexternal/gtest",
        "-Iexternal/gtest/include"
    ],
    linkopts = ["-pthread"],
    visibility = ["//visibility:public"],
)
```

Now `cc_` rules can depend on `//external:gtest/main`.

For example, we could create a test such as:

```cpp
#include "gtest/gtest.h"

TEST(FactorialTest, Negative) {
  EXPECT_EQ(1, 1);
}
```

Then create a BUILD file for your tests:

```python
cc_test(
    name = "my_test",
    srcs = ["my_test.cc"],
    copts = ["-Iexternal/gtest"],
    deps = ["@gtest//:main"],
)
```

You can then use `bazel test` to run the test.


Adding dependencies on precompiled libraries
--------------------------------------------

If you want to use a library that you only have a compiled version of (e.g.,
headers and a .so) wrap it in a `cc_library` rule:

```python
cc_library(
    name = "mylib",
    srcs = ["mylib.so"],
    hdrs = ["mylib.h"],
)
```

Then other C++ targets in your workspace can depend on this rule.