bundlewrap
bundlewrap copied to clipboard
bundlewrap
Config management with Python
See comment here: https://github.com/bundlewrap/bundlewrap/pull/829#pullrequestreview-2873916279
`/integration/bw_apply_git_deploy.py` leaves files behind that can't be removed without root permissions. ``` $ cd /tmp $ rm -Rf pytest-of-username/ rm: cannot remove 'pytest-of-username/garbage-bdcf595a-c12d-4ad8-b556-9106e6f6528b/test_deploy_from_url0/git_deployed_bw/.bundlewrap_git_deploy': Permission denied rm: cannot remove 'pytest-of-username/garbage-bdcf595a-c12d-4ad8-b556-9106e6f6528b/test_deploy_from_url0/git_deployed_bw/tests/unit/utils_text.py': Permission...
https://github.com/bundlewrap/bundlewrap/pull/825/files#diff-9c32388a3fbd66805995d1d52eb209bd21a6acef2b60ed19c82ad66bed95e553R8
Upstart has been dead for over 10 years: https://en.wikipedia.org/wiki/Upstart_(software) > As of June 2024, there have been no updates released for Upstart since September 2014. https://upstart.ubuntu.com/cookbook/ doesn’t even exist anymore.
Lets say you have this directory structure: ``` bundles/ foo/ items.py files/ a_file b_file ``` with `items.py` containing this: ```python files['/tmp/a_file'] = {} ``` The test should complain that `bundles/foo/files/b_file`...
When running `node_apply_end`-Hooks (and probably others), bundlewrap just prints `node_apply_end Running hooks from myhook.py`. Bundlewrap should show which node the hook is actually running on.
but just `__repr__()`, not the resolved fault, to avoid logging secrets to disk
``` ➤ bw metadata -k sla --blame routing-1 ╭──────┬───────────────────────────────────────────────────────────────────────────────────────────────────╮ │ path │ source │ ├──────┼───────────────────────────────────────────────────────────────────────────────────────────────────┤ │ sla │ group:hardware, group:group-a, group:routers, metadata_reactor:mybundle.set_sla │ ╰──────┴───────────────────────────────────────────────────────────────────────────────────────────────────╯ ``` I would like to have...
``` ➤ bw metadata -k sla --blame "lambda: not node.metadata.get('is_customer')" !!! blame information can only be shown for a single node ``` Why can't we have a table that shows...
the latter would make more sense since we want all attributes, not just the ones that have defaults, right?