Caliban

Latest version: v0.4.2

Safety actively analyzes 638681 Python packages for vulnerabilities to keep your Python projects secure.

Scan your dependencies

Page 2 of 5

0.2.3

- Added fix for an issue where large user IDs would crash Docker during the
build phase. https://github.com/google/caliban/pull/8

0.2.2

- Fix for bug with requirements.txt files.

0.2.1

- Added support for Conda dependencies
(https://github.com/google/caliban/pull/5). If you include `environment.yml`
in your project's folder Caliban will attempt to install all dependencies.
- Caliban now uses a slimmer base image for GPU mode.
- Base images for CPU and GPU modes now use Conda to manage the container's
virtual environment instead of `virtualenv`.

0.2.0

- Caliban now caches the service account key and ADC file; you should see faster
builds, BUT you might run into trouble if you try to run multiple Caliban
commands in the same directory in different processes, due to a race condition
with a temp file dropped in the directory. If you see a failure, try again.
- Private Cloud Source Repositories are now supported as requirements in the
`requirements.txt` and `setup.py` of projects executed using Caliban.
- `caliban notebook` now installs `jupyter` instead of `jupyterlab`. `caliban
notebook --lab` of course still uses `jupyterlab`.
- Caliban now works with Python 3.5.
- The default release channel for gke clusters in caliban is now 'regular', as
node autoprovisioning now works with preemptible instances in the regular
channel.
- If you provide `--cloud_key` Caliban will now properly use the supplied cloud
key to submit jobs to AI platform. Previously, Caliban would rely on the
system's auth method, which made it impossible to point to a different
project if your current service account key wasn't the owner.
- Changed gke job submission to accept min cpu and min mem arguments instead
of an explicit machine-type. This allows gke to efficiently schedule jobs
and prevents issues where jobs can be oversubscribed on compute nodes.
- added support for `.calibanconfig.json`. You can now add this file with an
`"apt_packages"` entry to specify aptitude packages to install inside the
container. The value under this key can be either a list, or a dictionary with
`"gpu"` and `"cpu"'` keys. For example, any of the following are valid:


This is a list by itself. Comments are fine, by the way.
{
"apt_packages": ["libsm6", "libxext6", "libxrender-dev"]
}


This works too:


You can also include a dictionary with different deps
for gpu and cpu modes. It's fine to leave either of these blank,
or not include it.
{
"apt_packages": {
"gpu": ["libsm6", "libxext6", "libxrender-dev"],
"cpu": ["some_other_package"]
}
}

0.1.15

- `caliban notebook` now attempts to search for the first free port instead of
failing due to an already-occupied port.

- `pip` is now called with `--no-cache-dir` inside the container; this should
shrink container sizes with no impact on performance.

- All commands have a new `--no-cache` option; when supplied, Docker will skip
using its build cache. This is helpful to use if you want to, say, force new
dependencies to get installed without bumping their versions explicitly.

0.1.14

- JSON experiment configuration files can now handle arguments which are varied
together, by supplying a compound key, of the form e.g. `[arg1,arg2]`.

- better error messages print when a docker command fails.

- Caliban can now handle pushing containers to "domain scoped projects":
https://cloud.google.com/container-registry/docs/overview#domain-scoped_projects
The colon in the project name separating domain and project ID is handled
properly.

Page 2 of 5

© 2024 Safety CLI Cybersecurity Inc. All Rights Reserved.