GitLab

Sunstudio could not find memset() in cstring; changed the include to
that of "string.h".  Also it complained of an anachronism in using
"extern int close()" as a default argument.  Removed these.

Conversion between socket structures is now done via the templated
convertSockAddr() function.

also constify some variables, and remove unnecessary cast.

Remove the READ/WRITE and DEFAULT macros by replacing them with
functions that thrown an exception on failure.  Other modifications
also made to make the code consistent with this model.

it's 0 or non 0).  same change for IPV6_V6ONLY, too.  also change ASSERT_EQs
to EXPECTs: failure of them doesn't seem to cause catastrophy in the subsequent
tests.  made also some style guideline fixes, too.

close the function if send_fd fails.

value identical to the parameter name itself.

Added the v6 only flag for IPv6 sockets. Both reuse address and this
flags get tested.

* Close the socket after sending from the creator, so it doesn't leak
  (but it doesn't seem to help :-().
* Correct range for the token numbers, so they are harder to guess
* Set the socket as reuse address

And for check_LTLIBRARIES to noinst_LTLIBRARIES too.

This is so a regular "make" will build the test programs and
libraries too.  So the "make check" will just run the tests (and
not build them).  This is to help with test-driven development.

This reverses some from #1091, plus does the rest too.

Note that doing a "make check" before a "make" does not currently
work due to dependencies depending on each other (e.g. python/isc/log
tests needs datasrc but python/isc/datasrc tests need log_config_update).

and while here also only build when the "make check" target is used;
since we already require --with-gtest maybe this step is not of much
interest.

For some reason, libtool is not picking up this library from the
dependency in libutil_unittests.  Adding the library to the link
cures the problem.

Also, move the directory util/io/tests to util/io_tests so that
the utilities code and tests build in the right order.

Also update the test Makefile.am files to include the appropriate libraries.

Since we couldn't have :: as destination address (it seems), we use ::1,
which should be possible.

- use <xxx.h> instead of <cxxx> (e.g. use stdlib.h instead of cstdlib)
- avoid using variable length array
as for the first point, it didn't make sense that cxxx didn't work and we may
want to look into it further to understand the real cause.  but since this has
been breaking build, and it somehow fixes the issue, I'll apply it for now.

okayed on jabber.

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3188 e5f2f494-b856-4b98-b285-d166d9295462

For one, many builtin functions do this (memcpy, read, etc). And, it
would be more correct to use unsigned char * (because of aliasing
rules), but that makes C++ unhappy when passing string literals there.

Got rid of -fno-strict-aliasing in sockcreator tests.

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3186 e5f2f494-b856-4b98-b285-d166d9295462

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3176 e5f2f494-b856-4b98-b285-d166d9295462

* Some of the code might be useful in other places, so it is split into
  libraries (isc::util::io and isc::util::unittests).
* Fixing flags in the sockcreator, leading to new warnings, fixing these
  warnings.
* TODO: read_data and write_data were tested implicitly with run(), but
  as they moved out to other library, they should have their own
  testcase.

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3163 e5f2f494-b856-4b98-b285-d166d9295462

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3162 e5f2f494-b856-4b98-b285-d166d9295462

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3161 e5f2f494-b856-4b98-b285-d166d9295462

The rest of code seems to have space between if/while and the
parenthesis.

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3160 e5f2f494-b856-4b98-b285-d166d9295462

And fixed a test. It seems underlaying bind does not like passing NULL
pointer as address and kills program with sigterm.

git-svn-id: svn://bind10.isc.org/svn/bind10/branches/vorner-sockcreator@3159 e5f2f494-b856-4b98-b285-d166d9295462