fuchsia: add base_paths_fuchsia.cc

base/path_service_unittest.cc

Index: base/path_service_unittest.cc
diff --git a/base/path_service_unittest.cc b/base/path_service_unittest.cc
index 599da000a9b3ba52e7cfca4f9716ba3c366fcb1e..d86bc0f92917f6f1db077d93865cbfb005cd948b 100644
--- a/base/path_service_unittest.cc
+++ b/base/path_service_unittest.cc
@@ -30,7 +30,7 @@ bool ReturnsValidPath(int dir_type) {
   // Some paths might not exist on some platforms in which case confirming
   // |result| is true and !path.empty() is the best we can do.
   bool check_path_exists = true;
-#if defined(OS_POSIX)
+#if defined(OS_POSIX) && !defined(OS_FUCHSIA)

[Lei Zhang, 2017/05/24 23:28:34]

Are we going to eventually testing Fuchsia here? Do we have a bug for that?

[scottmg, 2017/05/24 23:42:29]

This should test all paths that go through here, right? (Because it defaults to
check_path_exists = true).

I think that'll fail once we can actually link, but...

   // If chromium has never been started on this account, the cache path may not
   // exist.
   if (dir_type == DIR_CACHE)
@@ -115,7 +115,7 @@ TEST_F(PathServiceTest, Get) {
        ++key) {
     EXPECT_PRED1(ReturnsValidPath, key);
   }
-#elif defined(OS_POSIX)
+#elif defined(OS_POSIX) && !defined(OS_FUCHSIA)

[scottmg, 2017/05/24 23:42:29]

This part just avoids testing the POSIX_ specific bits, but we don't have any
FUCHSIA_ specific ones. The general ones are tested above in the first common
loop.

   for (int key = PATH_POSIX_START + 1; key < PATH_POSIX_END;
        ++key) {
     EXPECT_PRED1(ReturnsValidPath, key);