- moved b10-config_test.db from src/bin/stats/tests to
src/bin/stats/tests/testdata
- renamed its name to b10-config.db
- changed its format to multiline
These changes are due to the reviewer's comment.
Finally! This one was big.
With some minor updates to catch up to the changes after merge.
Conflicts:
src/bin/auth/auth_srv.cc
src/bin/auth/auth_srv.h
src/bin/auth/tests/auth_srv_unittest.cc
src/lib/datasrc/client_list.cc
src/lib/datasrc/tests/client_list_unittest.cc
When built from source it uses GTEST_HAS_PTHREAD.
See $GTEST_SOURCE/include/gtest/internal/gtest-port.h
which defines it for solaris, linux, hpux (which BIND 10 doesn't know yet).
In particular make it work with default --with-gtest-source which is "yes".
And also set the distcheck options the same (but not completely
tested due to #1910).
This is incomplete and not all tested.
Added --with-gtest-source=PATH for location of the googletest source.
Also tested old method (--with-gtest not source) with gtest 1.6.0.
The rationale for this is:
* Developers configure with `--enable-valgrind` and run `make check` to
see all the error reports (`make check` continues testing even if
there are Valgrind issues). This encourages us to fix these issues,
especially if we are modifying those parts of the code as part of
another bug.
* Builders configure with `--enable-valgrind
--enable-valgrind-suppressions` and run `make check` with
suppressions. `make check` doesn't show existing suppressed issues and
stops building if it finds a new Valgrind issue that is not
suppressed.
This should work fine with make check, so long as we have all issues
fixed or covered by suppressions. In this case, whenever a new issue
is introduced, make check would fail.
* Supply --enable-valgrind to configure to run tests (make check)
under valgrind
* Use two classes of suppressions files:
+ src/valgrind-suppressions for permanent suppressions that we
never want to re-consider again
+ src/valgrind-suppressions.revisit for temporary suppressions
(such as issues in testcases) that are not a priority to fix,
but we'd like to fix them sometime
This reverts commit 682df53277cfb545a1ba34a3a2198cb555cca6f5.
Decided to not enable documentation by default for now.
Need to handle build machines and make sure the doc generation
environment is in place first.
this is the real fix to the distcheck failure on some systems.
in the end LIBRARY_PATH_PLACEHOLDER didn't only unsolve the problem, but
also unnecessary at all. So the previous change on the makefile was
canceled completely.
