Scope macros containing var initialisations

[eelstork]

With some compilers declaring and initialising variables inside a switch may cause errors.
A real world example here.
Scoping these declarations solves the problem with hopefully no unwanted side effects.

[longjon]

Note that these should probably use do...while(0) blocks, see e.g. http://stackoverflow.com/questions/1067226/c-multi-line-macro-do-while0-vs-scope-block.

[eelstork]

@longjon is right. Updated.

[ukai]

build fails.
g++ -DHAVE_CONFIG_H -I. -I./src -I./src -pthread -Wall -Wwrite-strings -Wo
verloaded-virtual -Wno-sign-compare -DNO_FRAME_POINTER -g -O2 -MT src/logging_
unittest-logging_unittest.o -MD -MP -MF src/.deps/logging_unittest-logging_unitt
est.Tpo -c -o src/logging_unittest-logging_unittest.o test -f 'src/logging_unit test.cc' || echo './'src/logging_unittest.cc
In file included from src/utilities.h:82:0,
from src/logging_unittest.cc:33:
src/logging_unittest.cc: In function ‘void TestLogging(bool)’:
src/glog/logging.h:916:3: error: expected ‘;’ before ‘}’ token
}while(0)
^
src/glog/logging.h:949:3: note: in expansion of macro ‘SOME_KIND_OF_PLOG_EVERY_N
’
SOME_KIND_OF_PLOG_EVERY_N(severity, (n), google::LogMessage::SendToLog)
^
src/logging_unittest.cc:251:5: note: in expansion of macro ‘PLOG_EVERY_N’
PLOG_EVERY_N(ERROR, 2) << "Plog every 2, iteration " << COUNTER;
^
src/logging_unittest.cc:251:28: error: expected ‘;’ before ‘<<’ token
PLOG_EVERY_N(ERROR, 2) << "Plog every 2, iteration " << COUNTER;
^

[eelstork]

@ukai thanks for looking at this; would it be meaningful to setup some kind of C.I. to help with similar issues?
Working on this.

[eelstork]

@ukai since scoping macro contents breaks uses of the insertion operator, I suggest using intermediate variables so that anonymous scopes can be narrowed (demonstrated in latest commit, passes tests). Help is needed regarding proper naming and namespace inclusion of do_log and log_occurrences.

[ukai]

seems it makes LOG_EVERY macro thread unsafe?

[eelstork]

@ukai thanks for reviewing this PR. Yes, I can see how this may affect thread safety.