ExecContext's .sudo() omits many of run()'s parameters, and this
commit adds them. To avoid redundancy around repeating and massaging
the long parameter list of both functions and their return values, it
also adds some deeper changes:
- Make run(), _run(), sudo() and _sudo() always return instances of
Result. Before it was allowed to return a triplet of stdout,
stderr, and exit status.
- Have ExecContext stay out of the business of decoding the result
entirely. Result provides a convenience method .decode()
operating on stdout and stderr and leaves the decision to the
caller.
This entails miniscule adaptations in calling code, namely in
App.os_release, util.get_profile_env() and CmdListRepos._run().
- Wrap the _run() and _sudo() callbacks in a context manager object
of type CallContext to avoid code duplication.
- Consistently name the first argument to run(), _run(), sudo() and
_sudo() "cmd", not "args". The latter suggests that the caller is
omitting the executable, which is not the case.
Signed-off-by: Jan Lindemann <jan@janware.com>
Don't open and parse /etc/os-release with Python built-in functions.
Spawn "cat /etc/os-release" as a subprocess and capture the output
for parsing instead. The obvious advantage is that this also works
with a remote shell.
Signed-off-by: Jan Lindemann <jan@janware.com>
Use the AsyncRunner class introduced in the previous commit to add a
call_async() method, allowing to run async functions from sync
functions by spawning an extra event loop.
Signed-off-by: Jan Lindemann <jan@janware.com>
Add class AsyncRunner. This is a wrapper around the ceremony needed
to spawn an extra event loop in a synchronous function which wants to
call an async function.
Guido van Rossum considers it bad design that such a function exists
in the first place. While that may be true in the long run also for
jw-pkg, at this point I'm unwilling to flag every lazyly initialized
App property as async. It's not clear, yet, which will be async and
which not, and I dread the churn. So I will accept this as a
minimally invasive helper for now. When the API has stabilized, a
design without it may be better.
Signed-off-by: Jan Lindemann <jan@janware.com>
Expose App's __eloop member containing the application's main event
loop to allow outside async event loop trickery.
Signed-off-by: Jan Lindemann <jan@janware.com>
Take implementation burden from the derived classes _run() callback
by moving the respective code into the run() wrapper methods of the
base class.
Signed-off-by: Jan Lindemann <jan@janware.com>
Remove the key_filename parameter from the call to Paramiko's
connect(). It's user-dependent, and the current DevOps implementation
relies on having a SSH_AUTH_SOCK in the environment, anyway.
Signed-off-by: janware DevOps <devops@janware.com>
Add a wrapper around urlparse() and Paramiko's connect() function, in
order to log some more info in case an exception is thrown.
Signed-off-by: Jan Lindemann <jan@janware.com>
Take a positional uri argument to the constructor of ExecContext,
forcing SSHClient to follow suit. The latter was instantiated with a
hostname as only argument up to now, which still works as a special
case of an uri.
Signed-off-by: Jan Lindemann <jan@janware.com>
Use SSHClient as an ExecContext, i.e. use the .run() method instead
of .run_cmd(). Also, let SSHClient decide which implementation to
use.
Signed-off-by: Jan Lindemann <jan@janware.com>
Code beautification chore: Move the public methods of SSHClients to
the bottom of the class to be consistent with other classes.
Signed-off-by: Jan Lindemann <jan@janware.com>
Align the prototype of SSHClient.run_cmd() to ExecContext.run(). This
is a push towards making the SSHClient code an ExceContext, too. Some
arguments still log a warning or outright raise NotImplementedError.
Signed-off-by: Jan Lindemann <jan@janware.com>
CmdInfo._expand_macros() raises a custom exception during exception
handling. Replace that by logging some details and raising the
original exception to keep the stack trace readable.
Signed-off-by: Jan Lindemann <jan@janware.com>
Add the name of the operating system ID as taken from /etc/os-release
to $INSTALL_LOG. This should allow for easier post-mortem debugging
if multiple builds used the same file in /tmp and possibly also
prevent conflicts.
Signed-off-by: Jan Lindemann <jan@janware.com>
Make SSHClient accept a list of strings for the cmd argument to align
with the other run_cmd() functions in jw-pkg.
Signed-off-by: Jan Lindemann <jan@janware.com>
Move the PY_XXX = true|false variable definitions meant to be preset
by including makefiles to the top of py-defs.mk to make the structure
of the file clearer.
Signed-off-by: Jan Lindemann <jan@janware.com>
Add PY_INSTALL_INIT_PY ?= true to py-defs.mk. If set to false, a
Python module will not try to attempt installing an existing /
generated __init__.py. This is useful when installing into an exiting
directory with an existing __init__.py.
Signed-off-by: Jan Lindemann <jan@janware.com>
jw-pkg supports more than RPM-based package managers, but for
historic reasons, lots of its Makefile variables still have "RPM" in
their names. This is misleading. Replace "RPM" in variable names by
the more generic "PKG" where appropriate.
Signed-off-by: Jan Lindemann <jan@janware.com>
Remove stuff from rpmbuild.mk which is either unused, unusable or
not aligned with non-rpm-centric packaging workflows:
- Variable RPMBUILD
- Targets pkgbuild.dist pkg-upload-local.dist
- Variables used by these targets
Signed-off-by: Jan Lindemann <jan@janware.com>
Make all backend package manager prototypes have the same arguments:
yum(self, args: list[str], verbose: bool=True, sudo: bool=True)
This also implies having them behave equally verbose, unless
otherwise specified by the caller. This changes the default for
Debian.
Signed-off-by: Jan Lindemann <jan@janware.com>
Currently, the the version file is updated in the context of
pkg.sh release-reinstall -D "$(RPM_REQUIRES_DEVEL)"
RPM_REQUIRES_DEVEL is often filled from the current version, which in
turn is filled from the version file, so the order of events here is
unclear at best.
Add target pkg-release-update-version and make pkg-release-reinstall
depend on it to make the order explicit.
Signed-off-by: Jan Lindemann <jan@janware.com>
Make target pkg-release-reinstall depend on target get-official. It
already depends on get-maintainer, but that's not enough in
situations where devops built a target on platform A, pushed the new
release, then proceeds to build on platform B: It needs to pull its
own changes made during release of A.
Signed-off-by: Jan Lindemann <jan@janware.com>
make get-official already works as a pattern rule, but this commit
adds it explicitly to make tab-completion work.
Signed-off-by: Jan Lindemann <jan@janware.com>
pkg_relations_list() has an intricate case distinction around
expand_semver_revision_range, clean that up.
Signed-off-by: Jan Lindemann <jan@janware.com>
Use pkg-requires --hide-self to find all prerequisites that should be
installed for a test run against packages installed from the
repositories, including self-built and self-hosted packages.
Signed-off-by: Jan Lindemann <jan@janware.com>
In a push to eventually merge the classes, somewhat align the
command-line API of CmdRequiredOsPkg to the one of
BaseCmdPkgRelations by using dependency flavours as mandatory, first
argument.
Signed-off-by: Jan Lindemann <jan@janware.com>
To support the pkg-install-testbuild-deps target, a selector is
needed listing all prerequisites to be installed except the project
under test. --hide-self should be useful for that.
Signed-off-by: Jan Lindemann <jan@janware.com>
In a push to eventually merge class CmdRequiredOsPkg into this class,
add the --skip-excluded option required by it.
Signed-off-by: Jan Lindemann <jan@janware.com>
