A launcher that lets you configure software distributions and how they are consumed with dependency resolution. It provides a habitat for you to work in.
Features:
- URI based configuration resolution with inheritance. Makes it easy to define generic settings, but override for child URIs.
- Simple configuration with json files.
- Site configuration, code distributions are separate from URI configurations. All of these use a common json schema.
- Flexible site configuration with a minimum of environment variables.
- No long running python processes. The hab cli uses a shell specific script instead of a setuptools exe. This prevents some strange behavior in the shell if the python process is killed without exiting the new shell that python launched.
- Can modify the existing shell similar to how virtualenv's activate script works.
- Simple distribution of versions. Should not require updating of checked in files to release a unique distro version.
- Easy developer testing. A developer can additional site configurations for their host.
A git checkout can be found and the version of a distro can be dynamically generated
using setuptools_scm, or explicitly set by adding a
.hab_version.txtthat is not committed to the repo.
The general workflow for using hab is to use hab env to configure your shell, then
launch one or more aliases to work in. When done with that environment you exit it
with the shells exit command. Then you can enable another hab config with a
different URI.
You should be choosing the correct URI for your current task. This makes it so if required the underlying configuration for each task you are working on can be configured you don't need to know that something was changed.
$ hab env project_a/Seq001/S0010 # Enable this environment
[project_a/Seq001/S0010] $ maya # Launch a hab defined alias
[project_a/Seq001/S0010] $ houdini # Launch a hab defined alias
[project_a/Seq001/S0010] $ exit # Exit out of the current hab environment
$ # Back at the original shell without hab config
$ hab env project_a/Seq001/S0020 # Switch to another task with its own config ...C:\>hab env project_a/Seq001/S0010 # Enable this environment
[project_a/Seq001/S0010] C:\>maya # Launch a hab defined alias
[project_a/Seq001/S0010] C:\>houdini # Launch a hab defined alias
[project_a/Seq001/S0010] C:\>exit # Exit out of the current hab environment
C:\> # Back at the original shell without hab config
C:\>hab env project_a/Seq001/S0020 # Switch to another task with its own config ...PS C:\> hab env project_a/Seq001/S0010 # Enable this environment
[project_a/Seq001/S0010] C:\>maya # Launch a hab defined alias
[project_a/Seq001/S0010] C:\>houdini # Launch a hab defined alias
[project_a/Seq001/S0010] C:\>exit # Exit out of the current hab environment
PS C:\> # Back at the original shell without hab config
PS C:\> hab env project_a/Seq001/S0020 # Switch to another task with its own config ...In the above examples the user is enabling the hab environment for the
URI:project_a/Seq001/S0010, then launching the Maya and Houdini aliases.
Finally they are exiting the current URI so they can enable another URI. The user
doesn't need to worry about which version of the Maya or Houdini applications they
should launch, that is configured for them by the URI they pass to hab env.
You can enable tab completion including known URI's and aliases when using bash.
This requires using bash 4.4 or newer. To do this you need to source the
.hab-complete.bash file. This is installed next to the hab script (This makes
its path consistent even with editable installs). Use which .hab-complete.bash
to locate the file. If that doesn't work it should be next to the file returned
by which hab.
Examples of what to add to .bashrc
- Windows:
. /c/Program\ Files/Python39/Scripts/.hab-complete.bash - Linux:
. /usr/local/bin/.hab-complete.bash
See https://click.palletsprojects.com/en/8.1.x/shell-completion/ for more details.
In the previous section the use knew that they could run maya and houdini. You can
lookup the aliases for a given URI with the hab dump command. An alias is a hab
controlled way to launch a application.
$ hab dump project_a/Seq001/S0010
Dump of FlatConfig('project_a')
------------------------------------------------------
aliases: maya mayapy pip houdini houdini18.5
------------------------------------------------------This shows that a user has access to maya, mayapy, pip, houdini, and
houdini18.5 aliases. For this config the houdini and houdini18.5 aliases
end up launching the same application, See Multiple app versions
To support reusable alias shortcuts, hab has the ability to remember a URI and
reuse it for the next run. Anywhere you pass a URI to the cli, you can pass a
dash - instead. Hab will use the saved URI.
$ hab env -This feature has to be enabled by the site configuration. Depending on the site config
you may need to enable it by adding --prefs after hab.
hab --prefs dump -The site configuration may also configure a timeout that will require you to re-specify the URI after a while.
If you try to use - with an expired URI, Hab will notify you and prompt you to re-specify a URI.
To update the URI manually, pass --save-prefs after hab. You can
not use - when using this option.
hab --save-prefs dump project_a/Seq001/S0010User prefs are stored in the %LOCALAPPDATA% folder on Windows and in the user's
home directory on other platforms.
Hab is installed using pip. It requires python 3.6 or above. It's recommended
that you add the path to your python's bin or Scripts folder to the PATH
environment variable so you can simply run the hab command.
pip3 install hab
If you want to make use of json5 formatting when creating the various json files that drives hab, you should use the optional json5 dependency. This lets you add comments and allows for trailing commas.
pip3 install hab[json5]
Once hab is installed you need to point it to one or more site configurations using the HAB_PATHS environment variable. Each shell/platform assigns environment variables differently. These set the env variable for the current shell only.
export HAB_PATHS="/path/to/site_b.json:/path/to/site_a.json"You can use windows style paths by adding double quotes around the path(s). If
specifying a unc path with backslashes, you need to escape the leading slashes,
but can leave the remaining backslashes(when double quoted). Use ; for the pathsep.
export HAB_PATHS="c:\path\to\site_b.json;/c/path/to/site_a.json;\\\\server\share\path\to\site_a.json"set "HAB_PATHS=c:\path\to\site_b.json;/c/path/to/site_a.json"$env:HAB_PATHS="c:\path\to\site_b.json;c:\path\to\site_a.json"Hab has some code quality automated checks and extensive unit testing setup. When contributing, you should make sure your changes match the flake8 rules and follow the configured black formatting.
It's recommended that you install the configured pre-commit hooks. These test for and fix black and flake8 issues as well as other file checks.
Use pip to install pre-commit if it's not already installed.
pip3 install pre-commit
Install the hab configuration of pre-commit hooks to enable them when using git. Run this from the root of your git checkout.
pre-commit install
From this point forward when you commit changes using git, it will pre-validate and fix most code quality issues. Those changes are not staged so you can inspect the changes it makes.
We use tox to do all unit tests using pytest and also run black and flake8 code quality checks. You should use tox before you push your code changes to ensure that your tests will pass when running on github.
Install tox if its not already installed:
pip3 install tox
To run all tests. Make sure to run tox from the root of your git checkout:
tox
Examples of running specific tests:
tox -e py37-jsonRun just the py37-json testtox -e py39-json5Run just the py39-json5 testtox -e begin,py37-json,endShow code coverage report for just this testtox -e flake8Run the flake8 teststox -e begin,py37-json,end -- -vvEnables verbose mode for pytest. Any text after--is passed as cli arguments passed to pytest
identifier1/identifier2/...
You specify a configuration using a simple URI of identifiers separated by a /.
Currently hab only supports absolute URI's.
Examples:
- projectDummy/Sc001/S0001.00
- projectDummy/Sc001/S0001.00/Animation
- projectDummy/Thug
- default
If the provided URI has no configurations provided, the default configuration is used. This also supports inheritance with some special rules, see config inheritance for more details.
The provided URI is always stored in the HAB_URI environment variable.
Hab is designed as an api with cli support. The majority of the actual work is done by the api so it can be used with the provided cli, or can be customized by import. A gui version of the cli is planned in the future as a second pip package.
hab env: The env command launches a new shell configured by hab. You can exit the shell to return to the original configuration. This is how most users will interact with hab in the command line.hab activate: Updates the current shell with the hab configuration. This is similar to activating a virtualenv, but currently there is no way to deactivate. This is mostly how scripts can make use of hab.hab dump: Formats the resolved configuration and prints it. Used for debugging configurations and listing the commands available.hab launch: A shortcut forhab env --launch [alias], but automatically exits the configured shell when the launched application exits. Returning you to your existing shell without modification. Useful for quickly testing changes to a configuration requiring a running an application.hab set-uri: A command that allows the user to set a default URI. This default will be used with the dash"-"flag, which allows the user to quickly recall the saved URI. A argument can be passed to directly set the default. Alternatively, it can be run with no argument which will provide a prompt for the user to enter a URI. This method will display any currently saved URI for reference.
Examples:
$ hab env projectDummy
$ hab env projectDummy/ThugThe cli prompt is updated while inside a hab config is active. It is [URI] [cwd]
Where URI is the URI requested and cwd is the current working directory.
In most cases you will associate distros with URI's by specifying them in config files with the distros property.
There are times when you want to quickly load a new distro or force a distro to use other requirements. Here are a few reason why this might be useful.
- Change the version of a distro that is loaded temporarily.
- A plugin that only specific users would find useful, and others find annoying. For example department specific shelves in Maya/Houdini.
- Loading a plugin that requires an expensive license and only a few users will need.
- Adding a new distro to an existing URI when under development.
- Quickly testing a custom configuration of an existing URI.
- If you define a special empty URI that doesn't load any distros you can load any arbitrary set of distros to prototype a new URI.
A forced requirement must contain the name of the distro to use(the_dcc_plugin_a).
It can optionally contain requirement specifiers (the_dcc_plugin_a==1.0). When
used, the forced requirement replaces any distro requirements defined by the config.
This allows the forced requirement to replace an existing distro even if the forced requirement normally wouldn't be allowed. For this reason use it with caution. Optional distros provides you with a way to communicate what forced requirements are safe for users to use.
To use a forced requirement in the hab cli, use the -r or --requirement flag
before the COMMAND.
hab -r the_dcc_plugin_a dump -You can use it multiple times to load multiple distros:
hab -r the_dcc_plugin_a -r the_dcc==1.0 dump -Under normal operation hab resolves the configuration it loads each time it is run. This makes it easy to get updates to the configuration by re-launching hab. However, if you want to load the same hab configuration at a later date or on another workstation it's possible a new distro version has been released or some config settings have been modified. For example if you submit a render job to the farm, you want every frame to render using the same hab configuration not what ever it happens to resolve for that launch.
To handle this the hab cli stores the current configuration in the HAB_FREEZE
environment variable. You can even use a frozen config on other platforms as long
as you properly configure platform_path_maps in your site config.
The cli can be used to export these freezes. This example uses the cli to save a
freeze to disk as json using dump using --format json.
hab dump projectDummy --format freeze > /tmp/frozen_config.jsonAnd to restore that frozen config from the json file. This works for commands
other than dump.
hab dump --unfreeze /tmp/frozen_config.jsonSimilarly you can save/load the encoded freeze string using --format freeze.
This is what is stored in the HAB_FREEZE environment variable.
export frozen_config=$(hab dump app/nuke13 -f freeze)
hab dump --unfreeze $frozen_configA freeze string is prefixed with vX: to denote the version of freeze it was
encoded with. See hab.utils.encode_freeze and hab.utils.decode_freeze to
encode/decode each version of freeze strings.
You can configure what version of freeze string is saved in HAB_FREEZE by
setting the freeze_version key in your site json configuration. This should be
an int value or None. If not specified(ie None), then the default version is used.
{"set": {"freeze_version": 1}}These environment variables are used to configure hab.
| Name | Info |
|---|---|
| HAB_PATHS | Configure hab for the current computer. See site configuration files. |
| HAB_PYTHON | Configures the python used by the hab shell commands. See Python version. |
| HAB_RANDOM | Windows only, see Concurrency in Command Prompt. |
These environment variables are set by hab when it resolves a URI.
| Name | Info |
|---|---|
| HAB_FREEZE | A copy of the hab state used to activate the current session. This can be copied to another host and used to re-create the exact same state even across platforms. See restoring a resolved configuration. |
| HAB_URI | The URI used to activate the current session. |
Simple example of resolving an hab URI using the python api.
import hab
# Ensure the env var `HAB_PATHS` is set pointing to one or more site files.
resolver = hab.Resolver.instance()
# Load the config for the given URI
cfg = resolver.resolve("app/aliased")
# Launch the alias and run a command in a sub-process
proc = cfg.launch("as_str", ["-c", "print('done')"])The hab.Resolver.instance() call allows you to cache the configuration of all
configs and distros so they only need to be processed once. This allows isolated
code to access the same resolver information without needing a storage variable.
If you need more than one shared instance of Resolver you can pass name to
hab.Resolver.instance(name="studio"). You can also pass any kwargs that
hab.Resolver accepts, but they are ignored after the first call
(hab.Resolver.instance(name="studio", site=studio_site)).
The above examples are relying on the HAB_PATHS env var to define site. You can
also directly define a Site.
import hab
from pathlib import Path
root = Path(r'/mnt/studio/config')
site = hab.Site([root / "site_main.json", root / "site_override.json"])
resolver = hab.Resolver(site=site)Hab is configured by json files found with glob strings passed to the cli or defined by an environment variable.
The general rule for when hab encounters duplicated definitions is that the first encountered object is used and any duplicates are ignored. Here are some examples that follow that rule and links to details:
- Aliases: Multiple app versions
- Entry Points: Hab Entry Points
- Site: Site
config_paths and distro_paths have more complex rules.
Hab uses the HAB_PATHS environment variable to point to one or more site
configuration files. If the --site option is passed to the cli, it is used instead of
the environment variable.
Each of the file paths specified are read and merged into a single site configuration dictionary hab uses. When using multiple site json files here are some general rules to keep in mind.
- The left most site configuration takes precedence for a given item.
- For prepend/append operations on lists, the left site file's paths will placed on the outside of the the right site file's paths.
- For
platform_path_maps, only the first key is kept and any duplicates are discarded. - The entry_point
hab.site.add_pathsis processed separately afterHAB_PATHSor--sitepaths are processed, so:- Each path added is treated as left most when merging into the final configuration.
- The entry_point
hab.site.add_pathswill be ignored for dynamically added paths. - This can be used to include site files from inside of pip packages. For example a host installed pip package may be installed in the system python, or as a pip editable installation.
- Duplicate paths added dynamically are discarded keeping the first encountered(right most).
See Defining Environments for how to structure the json to prepend, append, set, unset values.
Developers can use this to load local site configurations loading their wip code instead of the official releases. See TestResolvePaths::test_paths to see an example of overriding the main site settings.
You can inspect the site settings by using the hab dump -t s or
hab dump --type site cli command. See
TestMultipleSites::test_left_right and TestMultipleSites::test_right_left
for an example of how these rules are applied. Here is a dump of the final result
of using all 3 site json files.
$ cd tests/site
$ hab --site site_left.json --site site_middle.json --site site_right.json dump --type site -v
Dump of Site
-------------------------------------------------------------------
HAB_PATHS: C:\blur\dev\hab_\tests\site\site_left.json
C:\blur\dev\hab_\tests\site\site_middle.json
C:\blur\dev\hab_\tests\site\site_right.json
config_paths:
distro_paths:
ignored_distros: release, pre
platforms: windows, osx, linux
set_value: left
test_paths: left_prepend
middle_prepend
right_prepend
right_append
middle_append
left_append
platform_path_maps: host: linux: host-linux_left
windows: host-windows_left
mid: linux: mid-linux_middle
windows: mid-windows_middle
net: linux: net-linux_right
windows: net-windows_right
shared: linux: shared-linux_left
windows: shared-windows_left
-------------------------------------------------------------------Note the order of left/middle/right in the test_paths variable. Also, for
platform_path_maps, host is defined in all 3 site files, but only the first
site file with it defined is used. The other path maps are picked up from the
site file they are defined in.
The site setting platform_path_maps is a dictionary, the key is a unique name
for each mapping, and value is a dictionary of leading directory paths for each platform.
PurePath.relative_to
is used to match, so full directory names need to be used. The unique name allows
for multiple site json files to override the setting as well as converting
resolved file paths to str.format style paths ({server-main}/folder/file.txt).
If multiple site json files specify the same key, the right-most site json file
specifying that key is used. It is safe to use forward slashes for windows paths.
{
"append": {
"platform_path_maps": {
"server-main": {
"linux": "/mnt/main",
"windows": "//example//main"
},
"server-dev": {
"linux": "/mnt/dev",
"windows": "//example//dev"
}
}
},
"set": {
"platforms": ["linux", "windows"]
}
}With these settings, if a path on a linux host, starts with /mnt/main when
generating the corresponding windows file path it will translate it to
\\example\main. Note the use of platforms to disable osx platform support.
The site file can be used to replace some hab functionality with custom plugins. These are defined in site files with the "entry_points" dictionary.
Example site json entry_point config.
{
"prepend": {
"entry_points": {
// Group
"hab.cli": {
// Name: Object Reference
"gui": "hab_test_entry_points:gui"
}
}
}
}See the Entry points specification data model for details on each item.
| Group | Description | **kwargs | Return | Multiple |
|---|---|---|---|---|
| hab.cli | Used by the hab cli to add extra commands. This is expected to be a click.command or click.group decorated function. |
All | ||
| hab.cfg.reduce.env | Used to make any modifications to a config after the global env is resolved but before aliases are resolved. | cfg |
All | |
| hab.cfg.reduce.finalize | Used to make any modifications to a config after aliases are resolved and just before the the config finishes reducing. | cfg |
All | |
| hab.habcache_cls | The class Site uses for its habcache features. This can be used to add features to the habcache system. | site |
First | |
| hab.launch_cls | Used as the default cls by hab.parsers.Config.launch() to launch aliases from inside of python. This should be a subclass of subprocess.Popen. A complex alias may override this per alias. Defaults to hab.launcher.Launcher. Example |
First | ||
| hab.site.add_paths | Dynamically prepends extra site configuration files to the current configuration. This entry_point is ignored for any configs added using this entry_point. | site |
A list of pathlib.Path for existing site .json files. |
All |
| hab.site.finalize | Used to modify site configuration files just before the site is fully initialized. | site |
All | |
| hab.uri.validate | Used to validate and modify a URI. If the URI is invalid, this should raise an exception. If the URI should be modified, then return the modified URI as a string. | resolver, uri |
Updated URI as string or None. | All |
The name of each entry point is used to de-duplicate results from multiple site json files. This follows the general rule defined in duplicate definitions.
Entry_point names should start with hab. and use . between each following word
following the group specification on https://packaging.python.org/en/latest/specifications/entry-points/#data-model.
When using multiple site config files you may end up needing to disable a entry point
in specific conditions. For example you want to enable
hab-gui's cli integration by default,
but need to disable it for linux farm nodes that have no gui enabled. If you set
a entry point's value to null hab will ignore it and not attempt to load it.
For example if you have your HAB_PATHS set to c:\hab\host.json;\\server\share\studio.json.
- The
host.jsonsite file is stored on each workstation and adds distros that are not able to be run from over the network. - The
studio.jsonsite file that is shared on the network for ease of deployment. It is used by all hab users on all platforms to define most of the studio defaults including adding the hab-gui cli as well as the URI configs.
// studio.json
{
"append": {
"entry_points": {
"cli": {
"gui": "hab_gui.cli:gui"
}
}
}
}For workstations that don't support a gui you modify the workstations host.json
site file, adding a override of the value from "hab_gui.cli:gui" to null.
// host.json
{
"append": {
"entry_points": {
"hab.cli": {
"gui": null
}
}
}
}Alternatively, you could create a second host site file named c:\hab\host_no_gui.json
put the gui disabling config in that file and on the host's you want to disable
the gui prepend to HAB_PATHS=c:\hab\host_no_gui.json;c:\hab\host.json;\\server\share\studio.json.
By default hab has to find and process all available configs and distros every
time it's launched. This has to glob the file paths in config_paths and
distro_paths, and parse each file it finds. As you add more distro versions
and configs this can slow down the launching of hab. This is especially true
when storing them on the network and when using windows.
To address this you can add per-site habcache files. This is a cross-platform
collection of all of the found files for a specific site file's config_paths
and distro_paths glob strings.
To enable caching run hab cache /path/to/site_file.json. This will create a
habcache file next to the site_file.json.
It will be named matching site["site_cache_file_template"][0], which defaults
to {stem}.habcache where stem is the site filename without extension. For the
example command it would create the file /path/to/site_file.habcache. To ensure
cross platform support, make sure your HAB_PATHS configuration contains all of
the required platform_path_maps site mappings.
While the site_file.habcache exists and HAB_PATHS includes site_file.json
hab will use the cached value unless the --no-cache flag is used. After adding,
updating or removing a config or distro, you will need to run the hab cache
command to update the cache with your changes. If using a distribution ci you
should add this command call there.
The habcache is cross platform as long as the hab site configuration loaded when
calling hab cache has all of the required platform_path_maps defined. The
cache will replace the start of file paths matching one of the current platform's
mapping values with the mappings key.
Hab uses shell script files instead of an entry_point executable. This allows it
to modify the existing shell(see hab activate). This has a small drawback of needing
to know what version of python to call. It relies on the assumption that you are using
hab with the default python 3 install. For example that you can call python3 -m hab
or py -3 -m hab on windows. Here is a breakdown of how the python call is built by
the scripts:
- If the
HAB_PYTHONenv var is set, its value is always used. - If a virtualenv is active, the
pythoncommand is used. - Otherwise on linux
python3is used, and on windowspy -3is used.
colorize: Ifhab dumpshould colorize its output for ease of reading.config_paths: Configures where URI configs are discovered. See below.distro_paths: Configures where distros discovered. See below.ignored_distros: Don't use distros that have this version number. This makes it possible for a ci to deploy some non-versioned copies of distros next to the distros so non-hab workflows can access known file paths. For example this could be used to put a latest folder next to each of the releases of a distro and not have to remove the .hab.json file in that folder.platform_path_maps: Configures mappings used to convert paths from one operating system to another. This is used by the freeze system to ensure that if unfrozen on another platform it will still work.platforms: A list of platforms that are supported by these hab configurations. When using freeze, all of these platforms will be stored. Defaults to linux, osx, windows.prereleases: If pre-release distros should be allowed. Works the same aspip install --pre ....prefs_default: Controls if user prefs can be used and if they are enabled by default. Ie if you can pass-for a URI. If this is set todisabled(the default), user prefs can not be used. If set to--prefs, then user prefs are enabled by default. Conversely, if set to--no-prefsthen user prefs are disabled by default. Users can pass either of these cli flags to hab to override the default(as long as its not disabled.)hab --prefs dump ....prefs_uri_timeout: If a URI user preference was set longer than this duration, force the user to re-save the URI returned for-when using the--save-prefsflag. To enable a timeout set this to a dictionary of kwargs to initialize adatetime.timedeltaobject.site_cache_file_template: The str.format template defining the name of habcache files.
config_paths and distro_paths take a list of glob paths. For a given glob
string in these variables you can not have duplicate values. For configs a
duplicate is two configs with the same URI. A duplicate distro is two distros
with the same name and version. If this happens a DuplicateJsonError is raised.
This prevents developers from copying a config and forgetting to update its context.
You can however have duplicates across individual glob paths. The glob paths are processed left to right and the first config/distro is used, any subsequent duplicates are ignored and a warning is logged to aid in debugging. This feature allows a developer to add any git checkouts they are working on that will be used, but still have access to all of the global shared configs/distros they are not working on. See specifying distro version for details on specifying a distro version in a git repo.
A distro defines a application, distribution or plugin that has multiple versions. It is mostly used to define aliases and environment variables. It can also define additional requirements.
A recommended released distro folder structure: [name]\[version]\.hab.json.
The [name] folder is referenced by one of the disto_path globs. This makes it easy
to store multiple versions of the distro. Each glob specified by distro_paths will
automatically have /*/.hab.json added to it, so the .hab.json file should
be in the root of a version folder. The root of the version folder is likely the root of
a git repo.
Example .hab.json:
{
"name": "maya2020",
"version": "2020.1",
"environment": {
"append": {
"MAYA_MODULE_PATH": "{relative_root}"
}
},
"aliases": {
"windows": [
["maya", "C:\\Program Files\\Autodesk\\Maya2020\\bin\\maya.exe"],
["mayapy", "C:\\Program Files\\Autodesk\\Maya2020\\bin\\mayapy.exe"]
]
}
}
In most cases you will not define version in .hab.json. If not defined, the
parent folder is used as the version. This makes it easy for automated deployments
without needing to modify a file checked into version control.
You will note that we are using the version of maya in name. This allows you to
provide access to multiple versions of the Maya application. Only one version of
a given distro name is going to be used so if you need access to multiple versions
of maya you must use this method. If an duplicate alias is defined, it is ignored.
The order distros are specified controls which duplicate alias is used, so make
sure the distro you want to use for a generic alias is specified before the others.
See app/houdini/a and
app/houdini/b for an example of how this
is controlled. Both of these configs end up adding the aliases houdini,
houdini18.5 and houdini19.5, but the houdini alias is configured differently.
There are a few ways to define a distro version, they are provided to make deployment and development testing easy. Here is the 4 ways to define the version of a distro, the first one found is used.
- The version property in
.hab.json. This has some drawbacks,.hab.jsonis likely checked into version control so modifying this requires committing changes to the repo, or working copy changes you have to maintain. - A
.hab_version.txtfile next to.hab.json. The drawback to this, is that it requires some maintenance to update, but allows you work around the issues from # 1 by not tracking this file in the repo. .hab.json's parent directory name. For distribution, this is the preferred option. You will end up needing a version folder for each deployed version of a disto to allow you to pick the version for a given config, so this lets you specify the version simply by copying it to the target location.setuptools_scm.get_versiongets a version from version control. This is for developer working copies, they can simply checkout the repo and even if its not a pip package this will resolve a valid and automatically updated version number provided the repo follows the setuptools_scm requirements for defining version numbers.
A config defines the environment to be applied. The context is picked by the provided URI. They are mostly used to define distros, pin those distros to specific versions, add alias_mods, and can be used to set environment variables.
A given config needs two pieces of information defined, its name and context. The context is a list of its parents names. When joined together they will build a URI.
Example project_a_thug_animation.json:
{
"name": "Animation",
"context": ["project_a", "Thug"],
"alias_mods": {
// Modify this env variable only when using the maya alias
"maya": {
"environment": {
"os_specific": true,
"windows": {"append": {"MAYA_MODULE_PATH": "//server/share/project_a"}}
}
}
},
"environment": {
"set": {
// Explicitly set this environment variable to a value
"STUDIO_PROJECT": "project_a"
}
},
// Inherit any configurations NotSet in this file from the parent context
"inherits": true,
// Require any version of the Maya2020 distro
"distros": [
"maya2020"
]
}This config would have the URI project_a/Thug/Animation.
Configs support min_verbosity with inheritance.
When doing bulk editing of multiple configs, the hab dump --type all-uris command
provides you with a easily diff-able json dump of the freeze
for all non-placeholder URI's defined. If a URI errors out when resolving, the
error text is stored instead of the freeze.
When resolving a URI it will find the closest exact match, so if project_a/Thug is
passed but Thug does not have a config, its parent project_a is used. If there is no
config for project_a, the default config will be used.
The config system has an inheritance system that follows a tree structure. If a property
is hab.NotSet on the chosen config and the config has inherit enabled the closest
parent with that property set will be used. If the root of the tree has inherit
enabled, and the property still is hab.NotSet, then the default tree will be
checked.
When the default tree is checked when resolving inheritance, some special rules for
matching contexts are applied. It will attempt to find the most specific context defined
in the default tree, but it will find the largest partial match for the start of each
URI identifier. In the default test config, you will see Sc1
and Sc11. The URI of not_a_project/Sc101 would end up using default/Sc1. The URI
not_a_project/Sc110 would use default/Sc11. The URI not_a_project/Sc200 would
use default.
The configuration environment variables and aliases can be formatted using
str.format. Hab extends the python Format Specification Mini-Language
with these extra features:
{ANYTHING!e}:!eis a special conversion flag for Environment variables. This will be replaced with the correct shell environment variable. For bash it becomes$ANYTHING, in power shell$env:ANYTHING, and in command prompt%ANYTHING%.ANYTHINGis the name of the environment variable.
Hab defines some specific variables. These are used when parsing an individual json file:
{relative_root}: The directory name of the .json config file. Think of this as the relative path.when using the command line, but this is a clear indication that it needs to be replaced with the dirname and not left alone.{;}: This is replaced with the path separator for the shell. Ie:for bash, and;on windows(including bash).
You can define custom variables in a json file by defining a "variables" dictionary.
You can use these keys in other parts of that file. An ReservedVariableNameError
will be raised if you try to replace hab specific variables like relative_root and ;.
{
"name": "maya2024",
"variables": {
"maya_root_linux": "/usr/autodesk/maya2024/bin",
"maya_root_windows": "C:/Program Files/Autodesk/Maya2024/bin"
},
"aliases": {
"linux": [
["maya", {"cmd": "{maya_root_linux}/maya"}]
],
"windows": [
["maya", {"cmd": "{maya_root_windows}\\maya.exe"}]
]
}
}The maya2024 testing example shows a way to centralize the path to the Maya bin directory. This way automated tools can easily change the install directory of maya if it needs to be installed into a custom location on specific workstations, or remote workstations.
Note: Using hard coded paths like maya_root_windows should be avoided unless
you really can't use relative_root. relative_root is more portable across
workstation setups as it doesn't require any modifications when the path changes.
Config and aliases can be hidden depending on the verbosity setting the user is using. The schema for this is:
"min_verbosity": {
"global": 1,
"hab": 2,
"hab-gui": 0
}The global key is used as a default value if a more specific key is requested
and not defined. Other keys can be used for more fine grained control of that
workflow. By default the target hab is used. The cli uses hab.
Filtering is not enabled by default, code needs to opt into it by using a with context.
import hab
# Specify the target when creating the Resolver instance(defaults to "hab").
resolver = hab.Resolver(target="something")
cfg = resolver.resolve('verbosity/inherit')
# By default, nothing will be hidden
print("a:", cfg.aliases.keys())
# Enable filtering by using the verbosity_filter with context.
with hab.utils.verbosity_filter(resolver, 0):
print("b:", cfg.aliases.keys())
with hab.utils.verbosity_filter(resolver, 1):
print("c:", cfg.aliases.keys())
Outputs:
a: dict_keys(['vb_default', 'vb0', 'vb1', 'vb2', 'vb3'])
b: dict_keys(['vb_default', 'vb0'])
c: dict_keys(['vb_default', 'vb0', 'vb1'])
If you use the target "hab-gui" instead of "something"
a: dict_keys(['vb_default', 'vb0', 'vb1', 'vb2', 'vb3'])
b: dict_keys(['vb_default', 'vb3'])
c: dict_keys(['vb_default', 'vb2', 'vb3'])
Aliases are used to run a program in a specific way. The hab cli creates shell commands for each alias. Aliases are defined on distro's per-platform.
{
"name": "aliased",
"aliases": {
"windows": [
[
"as_dict", {
"cmd": ["python", "{relative_root}/list_vars.py"],
"environment": {
"prepend": {
"ALIASED_GLOBAL_A": "Local A Prepend",
"ALIASED_LOCAL": "{relative_root}/test"
}
}
}
],
["as_list", ["python", "{relative_root}/list_vars.py"]],
["as_str", "python"]
]
},
"environment": {
"set": {
"ALIASED_GLOBAL_A": "Global A"
}
}
}This example distro shows the various ways you can define aliases. Each alias is defined as a two part list where the first item is the name of the created alias command. The second argument is the actual command to run and configuration definition.
Ultimately all alias definitions are turned into dictionaries
like as_dict, but you can also define aliases as lists of strings or a single
string. It's recommended that you use a list of strings for any commands that
require multiple arguments, for more details see args documentation in
subprocess.Popen.
When the aliases are converted to dicts, they automatically get a "distro" key
added to them. This contains a two item tuple containing the distro_name and
version. This is preserved when frozen. You can use this to track down what
distro created a duplicate alias. This is also useful
if you need to tie a farm job to a specific version of a dcc based on the active
hab setup.
import hab
import os
resolver = hab.Resolver.instance()
cfg = resolver.resolve(os.environ["HAB_URI"])
version = cfg.aliases["houdinicore"]["distro"][1]as_list and as_str show simple aliases. Ie aliases that just need to run a
command from inside the current hab environment. They inherit the environment
from the active hab config.
as_dict shows a complex alias, that also inherits the environment from the
active hab config, but in this case prepends an additional value on
ALIASED_GLOBAL_A. While this distro is in use, the environment variable
ALIASED_GLOBAL_A will be set to Global A. However while you are using the
alias as_dict, the variable will be set to Local A Prepend;Global A.
Complex Aliases supports several keys:
cmdis the command to run. When list or str defined aliases are resolved, their value is stored under this key.environment: A set of env var configuration options. For details on this format, see Defining Environments. This is not os_specific due to aliases already being defined per-platform.hab.launch_cls: If defined this entry_point is used instead of the Site defined or default class specifically for launching this alias. See houdini for an example.
Note: Plugins may add support for their own keys. Hab-gui adds icon and label for example.
Use Case: You want to add a custom AssetResolver to USD for Maya, Houdini,
and standalone usdview. To get this to work, you need to compile your plugin
against each of these applications unique compiling requirements. This means that
you also need to set the env var PXR_PLUGINPATH_NAME to a unique path for each
application. Maya's .mod files and houdini's plugin json files make it relatively
easy for the distro to update the global env vars MAYA_MODULE_PATH and
HOUDINI_PACKAGE_DIR to application specific configs setting the env var correctly.
However, that doesn't work for the standalone usdview application which doesn't
have a robust plugin loading system that can resolve application specific dll/so
files at startup. If you were to set the PXR_PLUGINPATH_NAME env var globally,
it would break houdini and maya as they would try to load the standalone path.
This is where complex aliases are useful. The usd distro can define an complex
alias to launch usdview that only adds the path to your standalone plugin to
PXR_PLUGINPATH_NAME only when that alias is launched.
Alias mods provide a way for a distro and config to modify another distro's aliases. This is useful for plugins that need to modify more than one host application independently.
{
"name": "aliased_mod",
"alias_mods": {
"as_list": {
"environment": {
"os_specific": true,
"windows": {
"set": {
"ALIASED_MOD_LOCAL_B": "Local Mod B"
}
}
}
}
},
}This example is forcing an the env var ALIASED_MOD_LOCAL_B to Local Mod B if
the resolved config has the alias as_list. Assuming the above aliased and this
aliased_mod distro are loaded, then the resulting as_list command would now set
the ALIASED_MOD_LOCAL_B env var to Local Mod B before calling its cmd.
Currently only the environment key is supported, and is os_specific.
Use Case: The complex alias use case has a drawback, it ties your usd version
to a specific usd plugin release. You likely will have multiple releases of your
plugin for the same usd version, as well as having multiple plugins you want to
version independently. To do this you can define a distro for each USD plugin,
and modify the PXR_PLUGINPATH_NAME env var for usdview using alias_mods.
You may find that you have to define more aliases than a regular user will regularly need, making the dump of aliases long and hard to parse. When using complex aliases, you can specify min_verbosity settings that will prevent showing some of those aliases unless you enable a higher verbosity.
For example all of site_main.json's aliases are:
maya mayapy maya20 mayapy20 pip houdini houdini18.5 houdinicore houdinicore18.5 husk husk18.5
While these are all useful, especially if you want to include access to
Multiple versions of maya or houdini, but in most cases
your users should only call maya or houdini, and don't need to call the script
aliases like mayapy, husk etc.
Using min_verbosity, you can make it so you have to increase the verbosity of dump's to see these advanced aliases. See .
By default just show the non-versioned aliases
$ hab dump default
Dump of FlatConfig('default')
-------------------------------------------
aliases: maya houdini houdinicore
-------------------------------------------
Adding a verbosity level now shows the version aliases as well
$ hab dump default -v
Dump of FlatConfig('default')
--------------------------------------------------------------------------
name: default
uri: default
aliases: maya maya20 houdini houdini18.5 houdinicore houdinicore18.5 husk
husk18.5
Adding another verbosity level now shows advanced aliases like mayapy and husk.
$ hab dump default -vv
Dump of FlatConfig('default')
-------------------------------------------------------------------------
name: default
uri: default
aliases: maya mayapy maya20 mayapy20 pip houdini houdini18.5 houdinicore
houdinicore18.5 husk husk18.5
The environment key in distro and config definitions is used to configure modifications
to the resolved environment. This is stored in HabBase.environment_config.
"environment": {
"unset": [
"UNSET_VARIABLE"
],
"set": {
"MAYA_MODULE_PATH": "{relative_root}"
},
"append": {
"MAYA_MODULE_PATH": "{relative_root}/append",
},
"prepend": {
"MAYA_MODULE_PATH": "prepend_value"
}
}There are 4 valid top level keys they are processed in this order if used:
- unset: The names of environment variables to remove.
- set: Replace or set the environment variable to this value.
- prepend: Treat this variable as a list and insert the value at the start of the list.
- append: Treat this variable as a list and add the value at the end of the list.
The unset key stores a list of environment variable names, the rest store a dictionary
of environment variable keys and the values to store.
The HabBase.environment property shows the final resolved
environment variables that will be applied. When using a resolved FlatConfig object,
environment also contains the merger of all environment_config definitions for all
distros. When building append and prepend environment variables it processes
each dependency in a depth-first manner.
These environment variables will be directly set if there is a value, and unset if the
value is blank. Hab doesn't inherit the session/system/user environment variable
values with the exception of the PATH variable as this would break the system.
Like Rez, the first set, prepend or append operation on a variable will replace the
existing variable value.
This quote from the Rez documentation explains why:
"Why does this happen? Consider PYTHONPATH - if an initial overwrite did not happen, then any modules visible on PYTHONPATH before the rez environment was configured would still be there. This would mean you may not have a properly configured environment. If your system PyQt were on PYTHONPATH for example, and you used rez-env to set a different PyQt version, an attempt to import it within the configured environment would still, incorrectly, import the system version."
If required, you can create OS specific environment variable definitions. To do this,
you nest the above structure into a dictionary with the correct windows or linux
key. You have to add a extra key os_specific set to true to indicate that you are
using os specific configurations.
"environment": {
"os_specific": true,
"windows": {
"append": {
"GOLAEM_WINDOWS_PATH": "C:\\Golaem\\Golaem-7.3.11\\Maya2020"
}
},
"linux": {
"append": {
"GOLAEM_LINUX_PATH": "/Golaem/Golaem-7.3.11/Maya2020"
}
}
}The distos key in distro and config definitions are used to define the distro version
requirements. When a config is processed the distro requirements are evaluated recursively
to include the requirements of the latest DistroVersion matching the specifier.
This uses the python packaging module
to resolve version specifiers so you can use the same configuration syntax you
would use in a pip requirements file.
"distros": [
"maya2020",
"maya2022",
"houdini18.5",
"hsite",
"animBot<=1.4",
"studiolibrary==2.5.7.post1",
// Use markers to only include 3ds max if hab is currently running on
// windows. It can't be run on linux.
"3dsmax2019;platform_system=='Windows'"
]The resolved versions matching the requested distros are shown in the versions property.
It also supports markers. Hab
should support all of the officially supported markers,
but the most common marker likely to be used with hab is platform_system.
The optional_distros key in config definitions are used to specify additional
distros some users may want to use. It allows you to easily communicate to users
on a per URI basis what additional distros they may want to use.
"optional_distros": {
"maya2024": ["Adds new aliases to launch Maya"],
"the_dcc==1.0": ["Optionally force the_dcc to use this version."],
"the_dcc_plugin_a": ["Load an optional plugin by default", true],
"the_dcc_plugin_b": ["Only have a few licenses for this plugin, so opt into loading it"]
}The dict key is the distro value that will be used. This can include version specifier
information like you see in the_dcc==1.0. See forced requirements.
The value is a list where the first item is a text description shown to the user to help them decide if they want to use this requirement. Optionally the second item is a bool that controls if interface plugins should enable this optional distro by default.
Optional distros are loaded using the forced requirements system. When using
the hab cli, the enabled by default option is not used, users will need to use
the -r/--requirement option.
Users can see the optional distros in the dump output with verbosity level of 1
or higher. hab dump - -v
Hab works on windows, linux and osx(needs tested). To make it easier to handle
platform specific code, it has all been moved into hab.utils.Platform instead
of directly relying on sys.platform, os.path, etc. This also has the
benefit of making it so the testing suite can test that hab works on for all
platforms without needing to test your code individually on each platform.
Ultimately we still need to test hab on each platform individually, but this
should help reduce surprises when the CI/CD runs the tests on a platform you
don't have easy access to when developing.
Hab can be forced to simulate being run on a different platform by replacing the Platform object.
hab.utils.Platform = hab.utils.WinPlatform
hab.utils.Platform = utils.BasePlatform.get_platform('osx')
# Restore the current platform
hab.utils.Platform = BasePlatform.get_platform()When working with tests, its recommended that you use the monkeypatch fixture.
monkeypatch.setattr(utils, "Platform", utils.LinuxPlatform)
monkeypatch.setattr(utils, "Platform", utils.WinPlatform)Hab doesn't use a console_script entry point to create an exe for its cli. It uses a shell
specific launch script. This script runs hab as a python process to create
temporary shell scripts to configure the shell(launching a new one if required).
This prevents the need to keep the python process running and prevents shell
corruption if that python process is killed. The shell scripts are written to the
temp location for the shell/os. On windows this should be written to %tmp% and
$TMPDIR on linux.
Hab does its best to remove these temp script files on exit so inspecting them can be
difficult. The best way to view them is to run a hab env or hab launch command
this will leave the hab process running while you find and view the config and launch
scripts in the temp directory. Once you are finished exit the hab process and they
will be removed. Alternatively if you use the --dump-scripts flag on hab commands
that write scripts, to make it print the contents of every file to the shell
instead of writing them to disk.
Hab uses python's logging module to output a ton of debugging information. For the
most part you can control the output using the hab -v ... verbosity option.
However if you need more fine grained control you can create a .hab_logging_prefs.json
file next to your user user prefs file. The cli also supports passing
the path to a configuration file using hab --logging-config [path/to/file.json]
that is used instead of the default file if pased.
- Using
hab activatein the command prompt is disabled. Batch doesn't have a function feature like the other shells, so each alias is its own .bat file in a directory prepended to the PATH environment variable. Hab needs to clean up its temp scripts and due to how activate works, there isn't currently a method to do this. We don't use doskey so we can support complex aliases. - Powershell disables running .ps1 scripts disabled by default. If you get a error like
hab.ps1 cannot be loaded because running scripts is disabled on this system., you will need to launch the Powershell with this commandpowershell -ExecutionPolicy Unrestricted. You can administratively default the execution policy to unrestricted for windows. - To use
hab activatein bash or Powershell you need to use.orsource. Powershell has the.operator so I would use that for both Powershell and bash.. hab activate default. - Jinja2 and MarkupSafe minimum requirements should be respected. This allows hab to work with Houdini 19.5 that ships with very dated versions of these packages. In practice this just means that we have to cast pathlib objects to strings before passing them to Jinja2.
This is the recommended way to launch a bunch of alias commands at once from within a python process.
import hab
# Resolve the hab configuration
hab_uri = 'app/aliased'
resolver = hab.Resolver.instance()
cfg = resolver.resolve(hab_uri)
# Launch the processes concurrently
procs = []
for _ in range(10):
proc = cfg.launch("as_str", ["-c", "print('done')"])
procs.append(proc)
# Example of waiting for all of them to finish and check for issues
errors = 0
for proc in procs:
out, _ = proc.communicate()
print(proc.returncode, '-' * 50, '\n', out)
if proc.returncode:
print("# BROKEN", proc.returncode)
errors += 1
print(f'Finished with {errors} errors')Using cfg.launch directly calls the alias with the correct environment variables
without the need for each subprocess to have to re-evaluate the hab environment,
then launch the actual alias application as its own subprocess.
If not using python, the preferred method is to use hab env to configure your
environment once, and then call the aliases in bulk.
This only applies to running hab in batch mode on windows using the cli and launching multiple processes at once using the same user. You may run into issues where the temp dir each hab command creates gets re-used by multiple processes.
By default hab.bat uses %RANDOM% which is very fast but uses time for its seed
which has a maximum resolution of seconds.
This is normally not an issue as a user is not able to call hab fast enough to
cause the issue.
A subdirectory or file C:\Users\username\AppData\Local\Temp\hab~7763 already exists.
If you end up seeing this message, or run into deleted file errors, then you can
set the env var HAB_RANDOM to one of these values. It is only respected when using
hab in batch mode.
| Value | Notes | ~ Time |
|---|---|---|
| fast | The default, uses %RANDOM% which is fast but may conflict when running concurrently. |
0.07s |
| safe | Uses python to generate a UUID. This is somewhat slower or it would be used by default. | 0.22s |
| [Anything else] | Anything else is a command to run and the output is captured as the unique value. |
Approximate time generated using time cmd.exe /c "hab -h" in git bash after
omitting the %py_exe% -m ... call.
- activate: Update the current process(shell) for a given configuration. Name taken from virtualenv.
- env: In the cli this launches a new process for a given configuration. Name taken from rez.
- config: Defines the environment variables and distros that should be used when a specific URI is requested.
- distro: Defines environment variables and aliases that a specific application or plugin requires, and other distros that it depends on.
- site: Apply specific settings to hab. Where to find distros and configs, etc.
- URI: A
/separated list of identifiers used to choose a specific config.
- Support per-alias environment variable manipulation. This will allow us to prepend to
PATH if required per-dcc. Ie only add
C:\Program Files\Chaos Group\V-Ray\3ds Max 2019\binto PATH, if and only if using launching 3ds Max any time its included in the distros. - Add support for
~and using environment variables in config and distro path resolution. - Add pkg_resource plugin interfaces that will allow customization. For example, how configurations are defined. Allowing the use of site specific database integrations etc.