Anaconda Ticket Workflow

A small Zsh productivity tool for repetitive Anaconda package-maintenance work. It keeps ticket workspaces and metadata together, remembers the active ticket, navigates to related feedstocks, and invokes the configured refinement and feedstock tools with their expected arguments.

The production workflow is entirely Zsh. Python is used only to test the real Zsh implementation in isolated temporary directories.

Features

• Create, register, edit, switch, list, and purge ticket workspaces.
• Restore the active ticket when a new shell sources the script.
• Navigate independently between ticket workspaces, the aggregate repository, and package feedstock repositories.
• Run the refinement and feedstock tools through stable wrapper commands.
• Store state in a small human-readable TSV registry.

Requirements

• Zsh.
• Conda initialized in the interactive shell when running internal tools.
• Access to the refinement and feedstock executables and the refinement support/configuration paths used by your team.
• Python 3.10 or newer and pytest for development and validation.

Installation

Clone or copy the repository to a stable location, then source the production script from .zshrc after the Conda initialization block:

source "$HOME/.config/anaconda-ticket-workflow/ticket-workflow.zsh"

Configuration exports must appear before that line. Restart Zsh or source the file directly after making changes:

zsh -n "$HOME/.config/anaconda-ticket-workflow/ticket-workflow.zsh"
source "$HOME/.config/anaconda-ticket-workflow/ticket-workflow.zsh"

Sourcing creates the configured ticket and state directories when necessary, then silently restores the last valid active ticket. It does not change the current directory, activate Conda, or invoke either internal tool.

Configuration

Environment variables are the public configuration interface. Set them before sourcing the script. Standard Zsh expressions such as $HOME expand when the export statement is evaluated; the workflow consumes the resulting value verbatim and never uses eval to reinterpret path text.

Variable	Required when	Default
TICKET_WORKFLOW_AGGREGATE_DIR	ticket init, cd_feedstock, or cd_aggregate	None
TICKET_WORKFLOW_TICKETS_DIR	Always available	$HOME/Workspace/tickets
TICKET_WORKFLOW_STATE_DIR	Always available	${XDG_STATE_HOME:-$HOME/.local/state}/anaconda-ticket-workflow
TICKET_WORKFLOW_REFINEMENT_TOOL	Refinement runs	None
TICKET_WORKFLOW_FEEDSTOCK_TOOL	Feedstock tooling runs	None
TICKET_WORKFLOW_REFINEMENT_SUPPORT_DIR	Refinement runs	None
TICKET_WORKFLOW_REFINEMENT_CONFIG	Refinement runs	None
TICKET_WORKFLOW_CONDA_ENV	Internal tools run	dev

For example:

export TICKET_WORKFLOW_AGGREGATE_DIR="$HOME/Workspace/aggregate"
export TICKET_WORKFLOW_TICKETS_DIR="$HOME/Workspace/tickets"
export TICKET_WORKFLOW_STATE_DIR="$HOME/.local/state/anaconda-ticket-workflow"
export TICKET_WORKFLOW_REFINEMENT_TOOL="/path/to/refinement-tool"
export TICKET_WORKFLOW_FEEDSTOCK_TOOL="/path/to/feedstock-tool"
export TICKET_WORKFLOW_REFINEMENT_SUPPORT_DIR="/path/to/refinement-support"
export TICKET_WORKFLOW_REFINEMENT_CONFIG="/path/to/refinement.ini"
export TICKET_WORKFLOW_CONDA_ENV="dev"

A copyable placeholder template is available at examples/ticket-workflow.env.example. The legacy runtime globals such as AGGREGATE_ROOT, TICKET_WORKSPACE_ROOT, and TICKET_STATE_DIR remain available for shell compatibility, but should not be used as the configuration interface.

For an existing installation, set the new variables to the locations already in use before sourcing this version. Registry records and state files are read without migration, and existing ticket directories are not moved. In particular, keep TICKET_WORKFLOW_TICKETS_DIR aligned with the parent directory recorded for existing ticket workspaces.

Command reference

Command	Purpose
ticket init <ticket-id> <package-name> <package-version>	Create and activate a ticket workspace, then run both tools. Must start in the aggregate directory.
ticket edit [<current-ticket-id>] [--ticket-id <new-id>] [--package-name <new-name>] [--version <new-version>]	Update a registry record and rename its ticket workspace. Without an ID, edit the active ticket.
ticket switch <ticket-id>	Make a registered ticket active without changing directory.
ticket list	List registered tickets and mark the active one.
ticket current	Show the active ticket and exported variables.
ticket run	Run both configured tools for the active ticket.
ticket purge	Permanently reset managed workspaces and state after typing PURGE.
ticket help, ticket -h, ticket --help	Show compact terminal help.

The current ticket ID for ticket edit, when supplied, must be the first argument after edit. Options cannot precede it.

Navigation and tool helpers

• cd_ticket [ticket-id] changes to a registered ticket workspace. An explicit ID does not change the active ticket.
• cd_feedstock [package-name] changes to <package-name>-feedstock under the aggregate. Without a name, it uses the active ticket's package.
• cd_aggregate changes to the configured aggregate repository.
• refinement_tool <ticket-id> <package-name> runs the configured refinement executable.
• feedstock_tool <ticket-id> runs the configured feedstock executable.

Ticket workspaces and feedstock repositories are separate. In particular, ticket edit only renames the managed ticket workspace. It never discovers, validates, renames, moves, or modifies a feedstock repository, and it does not activate Conda or run internal tools.

Typical workflow

The identifiers below are fictional placeholders:

cd_aggregate
ticket init EXAMPLE-100 configspace 1.2.2
ticket current
cd_ticket
cd_feedstock
ticket edit --version 1.2.3
ticket run
ticket switch EXAMPLE-101
ticket purge

Persistent state

The state directory contains:

tickets.tsv      # ticket ID, package, version, and workspace path
active-ticket    # active ticket ID
registry.lock    # temporary mutation lock; normally absent

The active record is exported as TI (ticket ID), PN (package name), PV (package version), and TD (ticket directory). State-only commands and cd_ticket do not require aggregate or internal-tool configuration.

Testing and development

Install the test dependency into the active development environment and run pytest:

python -m pip install -e '.[test]'
python -m pytest
zsh -n ticket-workflow.zsh

Tests launch real zsh -f subprocesses, source ticket-workflow.zsh, and use only temporary homes, state directories, workspaces, and mock executables. See CONTRIBUTING.md before changing public behavior.

Safety and limitations

• ticket purge permanently removes the configured ticket workspace and state roots. It rejects broad, overlapping, aggregate, symlinked, or inconsistent targets and requires exact interactive confirmation.
• Internal tool access and the expected Conda environment are organization- specific and are not included in this repository.
• ticket init registers and activates the workspace before running tools. A later tool failure does not remove the created ticket.
• Real interactive Ctrl+C behavior and integration with production internal tools may require manual verification.

Uninstall and cleanup

While the script is still loaded, run ticket purge if you intentionally want to delete all managed ticket workspaces and persistent state. Then remove the source line and configuration exports from .zshrc, start a new shell, and delete the copied repository. Skip ticket purge if the workspace data should be retained.