about_flags::ReportCustomFlags() should use signed 32-bit IDs.

chrome/browser/about_flags.cc

Index: chrome/browser/about_flags.cc
diff --git a/chrome/browser/about_flags.cc b/chrome/browser/about_flags.cc
index dd246e7703009eaed6fac106b2a59b162cbd0b7c..675fbf10671f7caf2702928f9e0bf2f126d6441f 100644
--- a/chrome/browser/about_flags.cc
+++ b/chrome/browser/about_flags.cc
@@ -2239,7 +2239,9 @@ void RecordUMAStatistics(FlagsStorage* flags_storage) {
 }
 
 uint32_t GetSwitchUMAId(const std::string& switch_name) {
-  return static_cast<uint32_t>(metrics::HashMetricName(switch_name));
+  // Need to strip sign bit.

[sky, 2014/08/27 20:24:53]

You need a better comment. Document 'why' we need to strip the sign bit, not
what the code is doing.

[Alexander Alekseev, 2014/08/28 02:48:55]

Done.

+  return static_cast<uint32_t>(metrics::HashMetricName(switch_name)) &

[Ilya Sherman, 2014/08/27 22:05:20]

Why not just static_cast to <int32_t>?  That has the added benefit that you'd
maintain consistency with the data that's currently being uploaded.

[Alexander Alekseev, 2014/08/28 02:48:55]

static_cast<int32_t>((uint64_t)(-1LL)) == -1

So I changed the return type, but AND is still needed here.

[Ilya Sherman, 2014/08/28 03:49:57]

I don't understand the problem.  What's wrong with reporting -1?

[Alexander Alekseev, 2014/08/28 14:40:52]

Except that it is unnatural, nothing wrong. Done.

PS: BTW, pretty-print.py formats comments inside enum with negative values
incorrectly. So this histograms.xml is uploaded with bypass-hooks option.

+         std::numeric_limits<int32_t>::max();
 }
 
 void ReportCustomFlags(const std::string& uma_histogram_hame,