Omitting an entrypoint.sh to configure the container and instead
depending on docker primitives allows us to be more portable. (If a
distribution uses a different mechanism for adding users, we need not
have multiple entrypoint.sh files or invariants within it; instead we
can configure that in the dockerfile itself along with all the other
distribution specific components.)
Ninja is not installed by default on the macOS machines; stop trying to
use it. Instead use `make -j` which should be roughly equivalent in
performance but supported everywhere.
Since we have migrated to Azure Pipelines, we have deprecated and
subsequentally removed all infrastructure for AppVeyor and
Travis. Thus it doesn't make a lot of sense to have the split
between "ci/" and "azure-pipelines/" directories anymoer, as
"azure-pipelines/" is essentially our only CI.
Move all CI scripts into the "azure-pipelines/" directory to have
everything centrally located and to remove clutter in the
top-level directory.
Right now, we have an awful hack in our test CI setup that extracts the
test command from CTest's output and then prepends the leak checker.
This is dependent on non-machine-parseable output from CMake and also
breaks on various ocassions, like for example when we have spaces in the
current path or when the path contains backslashes. Both conditions may
easily be triggered on Win32 systems, and in fact they do break our
Azure Pipelines builds.
Remove the awful hack in favour of a new CMake build option
"USE_LEAK_CHECKER". If specifying e.g. "-DUSE_LEAK_CHECKER=valgrind",
then we will set up all tests to be run under valgrind. Like this, we
can again simply execute ctest without needing to rely on evil sourcery.
Instead of having to find the fuzzer executables in our Azure test
scripts, provide test targets for each of our fuzzers that will
run them with the correct paths.
Our online tests are occasionally flaky since they hit real network
endpoints. Re-run them up to 5 times if they fail, to allow us to
avoid having to fail the whole build.
Our online tests are occasionally flaky since they hit real network
endpoints. Re-run them up to 5 times if they fail, to allow us to
avoid having to fail the whole build.
Subtle changes in the host OS can have impacts in the CI system that
may be hard to debug. We previously showed the results of `uname` which
can be difficult to interpret. Provide more information where
available.
Change the `GITTEST_REMOTE_PROXY_URL` environment variable to be
`GITTEST_REMOTE_PROXY_HOST`, since it is a host:port combination, not an
actual URL. (We cannot use a URL here since we may want to include the
username:password combination in the constructed URL.)
PowerShell can _read_ top-level variables in functions, but cannot _update_
top-level variables in functions unless they're explicitly prefixed with
`$global`.
Bind the proxy specifically to 127.0.0.1 instead of all addresses. This
is not strictly necessary for operations, but having a potentially open
proxy on a network is not a good idea.
Similar to the way we parse the ctest output on POSIX systems, do the
same on Windows. This allows us to append the `-r` flag to clar after
we've identified the command to run.
