Because of the fact that pkg-config is pants-on-head retarded and that
the Linux linker *requires* a static library to come before all its
dynamic dependencies in the link path, calling `pkg-config --libs
--static` was generating the wrong flags for linking.
Before this patch:
-Wl,-Bsymbolic-functions -Wl,-z,relro -L/usr/local/lib -lcurl -lssh2
-lrt -lgit2 -lssl -lcrypto -ldl -lz
After this patch:
-Wl,-Bsymbolic-functions -Wl,-z,relro -L/usr/local/lib -lgit2 -lcurl
-lssh2 -lrt -lssl -lcrypto -ldl -lz
By setting the "Libs" line before all other rules, we make sure that
`-lgit2` is the first library in the link path and that it gets its
symbols resolved with the libraries coming after it.
This fix (ab)uses an implementation detail in `pkg-config` (namely, that
flags are output as they are found on the file), but this detail seems
to be stable between releases and always gives a stable output.
These are treated as a list by CMake itself, which means that treating
them as a simple string can put semicolons in our ld command-line if we
have libraries which are not installed on the standard locations.
Treat the variable as a CMake list and replace it with the space-delimited
list just before writing it out to our pc file.
This reverts commit 969d4b703c.
This was a fluke from Coverity. The length to all the APIs in the
library is supposed to be passed in as nibbles, not bytes. Passing it as
bytes would prevent us from parsing uneven-sized SHA1 strings.
Also, the rest of the library was still using nibbles (including
revparse and the odb_prefix APIs), so this change was seriously breaking
things in unexpected ways. ^^
Pass on to whoever wants to link to libgit2 statically the flags that we
would have used for these libraries.
Putting them in Requires.private as we do now makes pkg-config put their
dependencies in the linker arguments as well, which is not what we want.
When diffing the index with the workdir and GIT_DIFF_UPDATE_INDEX has been passed,
the previous implementation was always writing the index to disk even if it wasn't
modified.
When a refspec contains no rhs and thus won't cause an explicit update,
we skip all the logic, but that means that we don't update FETCH_HEAD
with it, which is what the implicit rhs is.
Add another bit of logic which puts those remote heads in the list of
updates so we put them into FETCH_HEAD.
We currently recommend using `git_buf_grow` in order to make a buffer
make an owned copy of the memory it points to. This is not behaviour we
should encourage, so remove this recommendation.
The function itself is not changed, as we need to remain compatible, but
it will be changed not to allow usage on borrowed buffers.
When we don't own a buffer (asize=0) we currently allow the usage of
grow to copy the memory into a buffer we do own. This muddles the
meaning of grow, and lets us be a bit cavalier with ownership semantics.
Don't allow this any more. Usage of grow should be restricted to buffers
which we know own their own memory. If unsure, we must not attempt to
modify it.
