Index: build/linux/system.gyp |
diff --git a/build/linux/system.gyp b/build/linux/system.gyp |
index d9e553d7ddea8935317a9234238560295414765c..cbd2e59c4919d6753f0831496993faa6916dbbb6 100644 |
--- a/build/linux/system.gyp |
+++ b/build/linux/system.gyp |
@@ -480,6 +480,21 @@ |
'output_h': '<(SHARED_INTERMEDIATE_DIR)/library_loaders/libspeechd.h', |
'output_cc': '<(INTERMEDIATE_DIR)/libspeechd_loader.cc', |
'generator': '../../tools/generate_library_loader/generate_library_loader.py', |
+ |
+ # speech-dispatcher >= 0.8 installs libspeechd.h into |
+ # speech-dispatcher/libspeechd.h, whereas speech-dispatcher < 0.8 |
+ # puts libspeechd.h in the top-level include directory. |
+ # Since we need to support both cases for now, we ship a copy of |
+ # libspeechd.h in third_party/speech-dispatcher. If the user |
+ # prefers to link against the speech-dispatcher directly, the |
+ # `libspeechd_h_prefix' variable can be passed to gyp with a value |
+ # such as "speech-dispatcher/" that will be prepended to |
+ # "libspeechd.h" in the #include directive. |
+ # TODO(phaldan.jr): Once we do not need to support |
+ # speech-dispatcher < 0.8 we can get rid of all this (including |
+ # third_party/speech-dispatcher) and just include |
+ # speech-dispatcher/libspeechd.h unconditionally. |
+ 'libspeechd_h_prefix%': '', |
}, |
'action_name': 'generate_libspeechd_loader', |
'inputs': [ |
@@ -494,7 +509,9 @@ |
'--name', 'LibSpeechdLoader', |
'--output-h', '<(output_h)', |
'--output-cc', '<(output_cc)', |
- '--header', '<libspeechd.h>', |
+ '--header', '<<(libspeechd_h_prefix)libspeechd.h>', |
+ '--bundled-header', |
+ '"third_party/speech-dispatcher/libspeechd.h"', |
'--link-directly=<(linux_link_libspeechd)', |
'spd_open', |
'spd_say', |