Some implementation files were missing the license headers. This commit
adds them.

Some of our header files are not included at all by any of their
implementing counter-parts. Including them inside of these files leads
to some compile errors mostly due to unknown types because of missing
includes. But there's also one case where a declared function does not
match the implementation's prototype.

Fix all these errors by fixing up the prototype and adding missing
includes. This is preparatory work for fixing up missing includes in the
implementation files.

The zlib dependency includes "common.h" inside of the "zconf.h" header
to make available some type declarations like e.g. git_off_t. Including
the "common.h" header does pull in quite a lot of other headers though,
which are not required at all. Instead, we can just include our public
"git2/types.h" header, which is much more limited in its scope but still
provides everything required for "zconf.h".

This fix eases the transition later on to use a separate "features.h"
header instead of defines. As we have to generate the "features.h"
header, we put it in the build directory and add an include directory.
As we are splitting out building of dependencies into subdirectories,
this would mean that the zlib dependency needs to be aware of the parent
project's build directory, which is unfortunate. By including
"git2/types.h", we avoid this problem.

The current order of declarations and includes between "common.h" and
"w32_crtdbg_stacktrace.h" is rather complicated. Both header files make
use of things defined in the other one and are thus circularly dependent
on each other. This makes it currently impossible to compile the
"w32_crtdbg_stacktrace.c" file when including "common.h" inside of
"w32_crtdbg_stacktrace.h".

We can disentangle the mess by moving declaration of the inline crtdbg
functions into the "w32_crtdbg_stacktrace.h" file and adding additional
includes inside of it, such that all required functions are available to
it. This allows us to break the dependency cycle.

Instead of using INCLUDE_DIRECTORIES again for the libgit2_clar test
suite, we should just be using TARGET_INCLUDE_DIRECTORIES again if the
CMake version is greater than 2.8.11.

Apply `target_include_directories` when CMAKE_VERSION >= 2.8.12

Convert port with htons() in p_getaddrinfo()

Out of tree builds

cmake: Permit disabling external http-parser

`sin_port` should be in network byte order.

The test `refs::crashes::double_free` operates on our in-source
"testrepo.git" repository without creating a copy first. As the test
will try to create a new symbolic reference, this will fail when we want
to do a pure out-of-tree build with a read-only source tree.

Fix the issue by creating a sandbox first.

The test `index::tests::can_lock_index` operates on the "testrepo.git"
repository located inside of our source tree. While this is okay for
tests which do read-only operations on these resouces, this specific
test tries to lock the index by creating a lock. This will obviously
fail on out-of-tree builds with read-only source trees.

Fix the issue by creating a sandbox first.

Change the output path of generate.py to generate the clar.suite file
inside of the binary directory. This fixes out of tree builds with
read-only source trees as we now refrain from writing anything into the
source tree.

The source directory should usually not be touched when using
out-of-tree builds. But next to the previously fixed "clar.suite" file, we
are also writing the ".clarcache" into the project's source tree,
breaking the builds. Fix this by also honoring the output directory for
the ".clarcache" file.

The generate.py script will currently always write the generated
clar.suite file into the scanned directory, breaking out-of-tree builds
with read-only source directories. Fix this issue by adding another
option to allow overriding the output path of the generated file.

Our generate.py script is used to extract and write test suite
declarations into the clar.suite file. As is, the script accepts
multiple directories on the command line and will generate this file for
each of these directories.

The generate.py script will always write the clar.suite file into the
directory which is about to be scanned. This actually breaks
out-of-tree builds with libgit2, as the file will be generated in the
source tree instead of in the build tree. This is noticed especially in
the case where the source tree is mounted read-only, rendering us unable
to build unit tests.

Due to us accepting multiple paths which are to be scanned, it is not
trivial to fix though. The first solution which comes into mind would be
to re-create the directory hierarchy at a given output path or use
unique names for the clar.suite files, but this is rather cumbersome and
magical. The second and cleaner solution would be to fold all
directories into a single clar.suite file, but this would probably break
some use-cases.

Seeing that we do not ever pass multiple directories to generate.py, we
will now simply retire support for this. This allows us to later on
introduce an additional option to specify the path where the clar.suite
file will be generated at, defaulting to "clar.suite" inside of the
scanned directory.

When attempting to build libgit2 as an isolated static lib, CMake
gleefully attempts to use the system http-parser.  This is typically
seen on Linux systems which install header files with every package,
such as Gentoo.

Allow developers to forcibly disable using the system http-parser with
the config switch USE_EXT_HTTP_PARSER.  Defaults to ON to maintain
previous behavior.

Signed-off-by: Jason Cooper <jason@lakedaemon.net>

merge: fix potential free of uninitialized memory

The function `merge_diff_mark_similarity_exact` may error our early and,
when it does so, free the `ours_deletes_by_oid` and
`theirs_deletes_by_oid` variables. While the first one can never be
uninitialized due to the first call actually assigning to it, the second
variable can be freed without being initialized.

Fix the issue by initializing both variables to `NULL`.

merge: perform exact rename detection in linear time

travis: upgrade container to Ubuntu 14.04

Fix template dir empty string

Ubuntu 12.04 (Precise Pangolin) reaches end of life on April 28th, 2017.
As such, we should update our build infrastructure to use the next
available LTS release, which is Ubuntu 14.04 LTS (Trusty Tahr). Note
that Trusty is still considered beta quality on Travis. But considering
we are able to correctly build and test libgit2, this seems to be a
non-issue for us.

Switch over our default distribution to Trusty. As Precise still has
extended support for paying customers, add an additional job which
compiles libgit2 on the old release.

Some tests of ours require to be running against an SSH server.
Currently, we simply run against the SSH server provided and started by
Travis itself. As our Linux tests run in a sudo-less environment, we
have no control over its configuration and startup/shutdown procedure.
While this has been no problem until now, it will become a problem as
soon as we migrate over to newer Precise images, as the SSH server does
not have any host keys set up. Luckily, we can simply set up our own
unpriviledged SSH server. This has the benefit of us being able to
modify its configuration even in a sudo-less environment.

This commit sets up the unpriviledged SSH server on port 2222.

All our tests running against a local SSH server usually read the
server's URL from environment variables. But online::clone::ssh_cert
test fails to do so and instead always connects to
"ssh://localhost/foo". This assumption breaks whenever the SSH server is
not running on the standard port, e.g. when it is running as a user.

Fix the issue by using the URL provided by the environment.

adding GIT_FILTER_VERSION to GIT_FILTER_INIT as part of convention

travis: replace use of deprecated homebrew/dupes tap

Test improvements

Introduce a new test suite "odb::backend::simple", which utilizes the
fake backend to exercise the ODB abstraction layer. While such tests
already exist for the case where multiple backends are put together, no
direct testing for functionality with a single backend exist yet.

The fake backend currently implements all reading functions except for
the `exists_prefix` one. Implement it to enable further testing of the
ODB layer.

The `search_object` function takes the OID length as one of its
parameters, where its maximum length is `GIT_OID_HEXSZ`. The `exists`
function of the fake backend used `GIT_OID_RAWSZ` though, leading to
only the first half of the OID being used when finding the correct
object.

In order to be able to test the ODB prefix functions, we need to be able
to detect ambiguous prefixes in case multiple objects with the same
prefix exist in the fake ODB. Extend `search_object` to detect ambiguous
queries and have callers return its error code instead of always
returning `GIT_ENOTFOUND`.

Initialization of the `git_proxy_options` structure is never tested
anywhere. Include it in our usual initialization test in
"core::structinit::compare".