Introduce MinidumpStringWriter and its test

Introduce MinidumpStringWriter and its test.

MinidumpStringWriter is responsible for writing variable-length UTF-16
(MINIDUMP_STRING) and UTF-8 (MinidumpUTF8String) strings to minidump
files.

This change depends on https://codereview.chromium.org/430003003/ and
https://codereview.chromium.org/435013002/.

TEST=minidump_test MinidumpStringWriter.*
R=rsesek@chromium.org

Committed: https://chromium.googlesource.com/crashpad/crashpad/+/d2675ce50b7d

[Robert Sesek, 2014-08-01 18:27:29]

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
File minidump/minidump_string_writer.cc (right):

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer.cc:37: if (!MinidumpWritable::Freeze()) {
Not that it matters here, but I'd kind of expect to do local work prior to
calling the superclass implementation, otherwise you end up freezing children
before parents.

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer.cc:63: WritableIoVec iov;
I stared at this for a while and couldn't figure out why you write both the
string_base_ and string_ to the iovec.

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
File minidump/minidump_string_writer_test.cc (right):

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer_test.cc:67: };
nit: alignment

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer_test.cc:177: };
nit: alignment

[Mark Mentovai, 2014-08-01 18:33:58]

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
File minidump/minidump_string_writer.cc (right):

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer.cc:37: if (!MinidumpWritable::Freeze()) {
rsesek wrote:
> Not that it matters here, but I'd kind of expect to do local work prior to
> calling the superclass implementation, otherwise you end up freezing children
> before parents.

I think that it’s generally right to freeze the children before doing local
work, because parents might want to reach into their children to get stuff that
might not be nailed down until they’re frozen.

It doesn’t make a difference in this class, but it might in some of the more
complex ones.

[Robert Sesek, 2014-08-01 18:41:52]

LGTM

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
File minidump/minidump_string_writer.cc (right):

https://codereview.chromium.org/431363002/diff/1/minidump/minidump_string_wri...
minidump/minidump_string_writer.cc:37: if (!MinidumpWritable::Freeze()) {
On 2014/08/01 18:33:58, Mark Mentovai wrote:
> rsesek wrote:
> > Not that it matters here, but I'd kind of expect to do local work prior to
> > calling the superclass implementation, otherwise you end up freezing
children
> > before parents.
> 
> I think that it’s generally right to freeze the children before doing local
> work, because parents might want to reach into their children to get stuff
that
> might not be nailed down until they’re frozen.
> 
> It doesn’t make a difference in this class, but it might in some of the more
> complex ones.

Yeah, I figured it could go either way. I didn't know if there was a reasoning
for doing the children before or after.

[Mark Mentovai, 2014-08-01 18:45:00]

Committed patchset #3 manually as d2675ce50b7d (presubmit successful).