A few things that don't work or aren't tested yet:
- wrappers at the top level
- equality checking for not-yet-expanded wrappers.

This fixes build on, at least, NetBSD 8.1 with gcc 5.5 and SmartOS with gcc 7.

Java: Update to Guava 28.1-android

Fixed bug in Ruby DSL when no names are defined in a file.

Ported names fix for Ruby to the release branch.

Merge 3.10.x with master and increment the version number

* Fix conformance test

Default value of string/message map is not encoded

* Fix zts build

Update CHANGES.txt with changes since 3.10.0-RC1

* Initialize well known type values

Also fixes #6667

* Revert unwanted change

* Initialize well known type values

Also fixes #6667

* Revert unwanted change

This reverts commit 911e3040fa597e6a82351375d9eb416a6d3fae3a.

Currently, the generated header file might cause warnings if user specifies `-Wconversion` compiler option. This commit should be able to fix it.

* Update README.md

Fixed spelling error

* Update README.md

When browsing around the strutil files I found a function
that was never referenced inside the code base
"void StripString(string* s, const char* remove,
-                                 char replacewith);"

The name was kind of misleading as well and it seems like
it's a carbon copy of
"void ReplaceCharacters(string* s, const char* remove,
                                       char replacewith);"
(even the parameter names are the same, the code is the same..)

Is it intentional? Maybe for compatibility reasons? If so,
let's make it deprecated and use the ReplaceCharacters method inside
or the other way around.

Also, noticed there were no tests for "StripString" or "Replace".
Added some for both and planning on maybe making it more C++ish (?)
in another commit.

* Add document for php development

* Fix comments

Update Java Readme to 3.9.2

3.9.2 is on the Githib releases page.

The new function name also better reflects the semantics of the
function.  Like timegm(), this function always converts to/from
UTC, not local time.

Pull the ObjC fixes from #6679 to the 3.10.x branch

The OSReadLittleInt64 function as defined by Apple reduces down to:

`return *(volatile uint64_t *)((uintptr_t)base + byteOffset);`

which means we are type-punning using a cast. On ARMv7 and other aligned architectures this can cause crashes. 
Minimal example: https://gist.github.com/dmaclach/b10b0a71ae614d304c067cb9bd264336

Fixes #6679 

The code in question hasn't change in a long time so the cause of
https://github.com/firebase/firebase-ios-sdk/issues/3851 still appears to be
an Xcode 11 clang change/bug; but this does appear to be slightly better
code for the work being done.

Cleanup along the way for #6679 

The OSReadLittleInt64 function as defined by Apple reduces down to:

`return *(volatile uint64_t *)((uintptr_t)base + byteOffset);`

which means we are type-punning using a cast. On ARMv7 and other aligned architectures this can cause crashes. 
Minimal example: https://gist.github.com/dmaclach/b10b0a71ae614d304c067cb9bd264336

Fixes #6679