clang tools: Add --bootstrap to the documentation for update.py

tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp

Index: tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp
diff --git a/tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp b/tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp
index db208abc2f762e0a7848ddb5f3700cb21743c46f..08ce7b7871d365c8a0fd54abea57029fba727aa8 100644
--- a/tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp
+++ b/tools/clang/rewrite_to_chrome_style/RewriteToChromeStyle.cpp
@@ -111,7 +111,10 @@ bool GetNameForDecl(const clang::FieldDecl& decl,
       original_name.substr(strlen(kBlinkFieldPrefix)));
   // The few examples I could find used struct-style naming with no `_` suffix
   // for unions.
-  if (decl.getParent()->isClass())
+  bool c = decl.getParent()->isClass();
+  // There appears to be a GCC bug that makes this branch incorrectly if we
+  // don't use a temp variable!! Clang works right. crbug.com/580745
+  if (c)

[dcheng, 2016/01/27 23:46:15]

This gives me the heebie jeebies. I feel like we shouldn't check in code like
this, because if we can't trust the compiler, then there's basically nothing we
can do, right? We basically don't have a way to guarantee that this will work
since it's really hard to test.

[danakj, 2016/01/27 23:51:01]

Ya, I'm sure there's other bugs happening elsewhere too, heh. I just don't like
leaving the one I know here malfunctioning, just cuz.. you think it's better to
cover our eyes on this one place and just say don't use gcc?

     name += '_';
   return true;
 }