Android: Bump up the fallback OS version number to 4.3.0

base/sys_info_android.cc

 // Copyright (c) 2012 The Chromium Authors. All rights reserved.
 // Use of this source code is governed by a BSD-style license that can be
 // found in the LICENSE file.
 
 #include "base/sys_info.h"
 
 #include <sys/system_properties.h>
 
 #include "base/logging.h"
 #include "base/strings/string_number_conversions.h"
 #include "base/strings/string_piece.h"
 #include "base/strings/stringprintf.h"
 
 namespace {
 
 // Default version of Android to fall back to when actual version numbers
 // cannot be acquired.
 // TODO(dfalcantara): Keep this reasonably up to date with the latest publicly
 //                    available version of Android.
-static const int kDefaultAndroidMajorVersion = 4;
-static const int kDefaultAndroidMinorVersion = 0;
-static const int kDefaultAndroidBugfixVersion = 3;
+const int kDefaultAndroidMajorVersion = 4;
+const int kDefaultAndroidMinorVersion = 3;

[jar (doing other things), 2013/07/27 18:18:01]

Why isn't this stuff pulled from the OS?

This seems to bake into Chrome the version number for the OS... but yet we could
be running on any version of the OS.

What am I missing?

[joth, 2013/07/27 19:23:39]

The clue is in "kDefault..." (and, in the comment above it) - this is only used
if the real OS version can't be sucked out of the OS at runtime. (See the method
immediately below). This should only fail when the app is running on
pre-production HW / OS that does not yet have an OS version assigned. We
fallback to a sane known-released OS version.
Letting this get too old (or report project code-names for that matter) means it
falls foul of UA sniffing on sites that trigger legacy support when the OS is
not known (or at least,  can lead people to incorrectly suspect that maybe
happening in weird bug diagnosis, which is was what happened to me yesterday)

[jar (doing other things), 2013/07/28 19:19:02]

That is mostly a great answer... but...

Is it really preferable to conflate results for a known OS with trash when the
OS is not responding with its real version?  I would think it would be desirable
to have a very different (nonexistent??) OS to handle this case, rather than
merge in with a real version.

Can you explain why this is good?

[jar (doing other things), 2013/07/29 19:03:27]

IMO, sending up the old answer will seemingly allow us to spot when we are not
getting data (not getting the real OS version).... unless we'll filter these out
completely <sigh>.

+const int kDefaultAndroidBugfixVersion = 0;
 
 // Parse out the OS version numbers from the system properties.
 void ParseOSVersionNumbers(const char* os_version_str,
                            int32 *major_version,
                            int32 *minor_version,
                            int32 *bugfix_version) {
   if (os_version_str[0]) {
     // Try to parse out the version numbers from the string.
     int num_read = sscanf(os_version_str, "%d.%d.%d", major_version,
                           minor_version, bugfix_version);
@@ skipping 125 matching lines @@
   return heap_size;
 }
 
 int SysInfo::DalvikHeapGrowthLimitMB() {
   static int heap_growth_limit = GetDalvikHeapGrowthLimitMB();
   return heap_growth_limit;
 }
 
 
 }  // namespace base