Fix "conversion of size_t to unsigned int" warning

net/dns/record_rdata.h

Index: net/dns/record_rdata.h
diff --git a/net/dns/record_rdata.h b/net/dns/record_rdata.h
index d2ca587c5e7811ebb72ab8fce6db6f1c854acc90..7b4bddf301c3b2c01d013b67db1eadce0f281029 100644
--- a/net/dns/record_rdata.h
+++ b/net/dns/record_rdata.h
@@ -198,7 +198,11 @@ class NET_EXPORT_PRIVATE NsecRecordRdata : public RecordRdata {
   uint16_t Type() const override;
 
   // Length of the bitmap in bits.
-  unsigned bitmap_length() const { return bitmap_.size() * 8; }
+  // This will be between 8-256, as the bitmap is required to be between 1-32
+  // bytes long (this is verified by Create()).

[Ryan Sleevi, 2016/07/06 16:29:03]

This is the sort of comment that makes me uneasy in the header (the
parenthetical), because that talks about implementation.

// This will be between 8 and 256 bits long, per RFC [blah] Section [blah]

[Rob Percival, 2016/07/06 17:14:44]

I had the same feeling when I wrote it. I wasn't convinced just referring to the
RFC would be sufficient though, since what the RFC allows and what comes over
the network can be very different. I wanted to say something closer to what the
DCHECK was intended to express - that it is an internal guarantee.

+  uint16_t bitmap_length() const {
+    return static_cast<uint16_t>(bitmap_.size() * 8);
+  }
 
   // Returns bit i-th bit in the bitmap, where bits withing a byte are organized
   // most to least significant. If it is set, a record with rrtype i exists for