Fix Cronet compilation error due to different kCharsetLatin1 definition.

net/base/net_string_util_icu_alternatives_android.cc

Index: net/base/net_string_util_icu_alternatives_android.cc
diff --git a/net/base/net_string_util_icu_alternatives_android.cc b/net/base/net_string_util_icu_alternatives_android.cc
index f4ef1efc81a12c4f4002bf4f78bea4f2a8bde283..ca0e7f0315db7677a924b825495b0d27fbdb2df3 100644
--- a/net/base/net_string_util_icu_alternatives_android.cc
+++ b/net/base/net_string_util_icu_alternatives_android.cc
@@ -66,7 +66,7 @@ ScopedJavaLocalRef<jstring> ConvertToJstringWithSubstitutions(
 
 }  // namespace
 
-const char kCharsetLatin1[] = "ISO-8859-1";
+const char* const kCharsetLatin1 = "ISO-8859-1";

[xunjieli, 2015/01/28 15:54:57]

I think crrev.com/696033002 was trying to change char* to  char[]. The latter
seems to be the recommended style, so we should probably not change it back to
char*. How about change the declaration in net_string_util.h to char[] and the
definition in net/base/net_string_util_icu.cc to char[] as well?

[mef, 2015/01/28 19:06:59]

Unfortunately changing net_string_util_icu to 

const char kCharsetLatin1[] = base::kCodepageLatin1;

gets an error 

../../net/base/net_string_util_icu.cc:14:12: error: array initializer must be an
initializer list or string literal

as base::kCodepageLatin1 is also const char [].

 

[xunjieli, 2015/01/28 19:15:22]

Ahh, i see. Maybe add a comment here so it won't get changed to char[] in future
cleanup?

[mmenke, 2015/01/30 15:28:23]

SGTM

[mef, 2015/01/30 15:38:27]

Done.

 
 bool ConvertToUtf8(const std::string& text, const char* charset,
                    std::string* output) {