Move most importer code to chrome/utility/importer

chrome/browser/importer/in_process_importer_bridge.cc

Index: chrome/browser/importer/in_process_importer_bridge.cc
diff --git a/chrome/browser/importer/in_process_importer_bridge.cc b/chrome/browser/importer/in_process_importer_bridge.cc
index 61d09d21ea531b13e04ae1679c9519470bd7f571..742deb7d6971f6cb83d56d01568a13caf4358552 100644
--- a/chrome/browser/importer/in_process_importer_bridge.cc
+++ b/chrome/browser/importer/in_process_importer_bridge.cc
@@ -54,7 +54,7 @@ history::VisitSource ConvertImporterVisitSourceToHistoryVisitSource(
       return history::SOURCE_IE_IMPORTED;
     case importer::VISIT_SOURCE_SAFARI_IMPORTED:
       return history::SOURCE_SAFARI_IMPORTED;
-  }
+  };
   NOTREACHED();

[gab, 2013/07/09 20:48:31]

I think these last 2 lines can be removed.

[scottmg, 2013/07/09 21:16:45]

I think they should be kept in case new enums are added. I can put them in a
default: if you prefer.

The ; can definitely be nuked!

[gab, 2013/07/09 21:29:43]

Well if they are removed and new enums are added, this won't compile, if we
leave them there, then it will compile, but not behave properly...

I prefer the former :).

[scottmg, 2013/07/09 23:03:09]

I don't believe that's true. e.g. win vs2010 release:

d:\src\cr3\src\chrome\browser\importer\in_process_importer_bridge.cc(58) :
warning C4715: '`anonymous
namespace'::ConvertImporterVisitSourceToHistoryVisitSource' : not all control
paths return a value

I can see a reasonable argument either way for having it inside or outside of
the switch though.

   return history::SOURCE_SYNCED;
 }