taskcheck 0.3.0

Automatic scheduler for taskwarrior

A non-AI automatic scheduler for taskwarrior (i.e. alternative to skedpal/timehero/flowsavvy/reclaim/trevor/motion)

This is a taskwarrior extension checks if tasks can be completed on time, considering estimated time and working hours.

Features

• Use arbitrarily complex time maps
• Use ical to block time from scheduling (e.g. for meetings, vacations, etc.)
• Implement scheduling algorithm for parallely working on multiple tasks
• Use Google API to access calendars
• Export tasks to iCal calendar and API calendars

Install

1. pipx install taskcheck
2. taskcheck --install

How does it work

This extension parses your pending and waiting tasks sorted decreasingly by urgency and tries to schedule them in the future. It considers their estimated time to schedule all tasks starting from the most urgent one.

UDAs

Taskcheck leverages to UDAs, estimated and time_map. The estimated attribute is the expected time to complete the task in hours. The time_map is a comma-separated list of strings that indicates the hours per day in which you will work on a task (e.g. work, weekend, etc.). The exact correspondence between the time_map and the hours of the day is defined in the configuration file of taskcheck. For instance:

[time_maps]
# get an error)
[time_maps.work]
monday = [[9, 12.30], [14, 17]]
tuesday = [[9, 12.30], [14, 17]]
# ...

They say it's an "AI"

Taskcheck will also parse online iCal calendars (Google, Apple, etc.) and will match them with your time maps. It will then modify the Taskwarrior tasks by adding the completion_date attribute with the expected date of completion and the scheduled attribute with the date in which the task is expected to start.

It will also print a red line for every task whose completion_date is after its due_date.

In general, it is recommended to run taskcheck rather frequently and at least once at the beginning of your working day.

Reports

You can also print simple reports that exploit the scheduling UDA filled by Taskcheck to grasp how much time you have to work on which task in which day. For instance:

• taskcheck -r today will show the tasks planned for today
• taskcheck -r 1w will show the tasks planned for the next week,

Algorithms

Taskcheck provides two scheduling algorithms: parallel and sequential.

Sequential

The sequential algorithm schedules tasks one after the other. It is the simplest algorithm. It looks at your tasks one by one, computes the time slots available for that task, and fills them as soon as possible.

While it sounds logical, it has a problem. While you're working on a task, its remaining estimated time is decreasing, and if your urgency depends on the estimated time left, the urgency of the task will decrease as well. This is especially problematic for long tasks: imagine if you block your whole week for a single task and all the other small tasks, even if urgent, don't get scheduled. After a few days, the urgency of the first long task should have decreased.

Parallel

The parallel algorithm tries to solve the problem of the sequential algorithm.

Instead of allocating all the available time for a task until the task is finished, the parallel algorithm allocates a fixed amount of time for each task (e.g. 1 hour) and then recomputes the urgency and restart the allocation procedure.

If after 1 hour the long task has decreased urgency, that will be noticed and the newer most urgent task will get scheduled in its place.

The minimum time for a task is by default 2 hours, but you can change it by tuning the min_block taskwarrior UDA.

The parallel algorithm is the default algorithm.

Configuration

taskcheck --install allows you to create required and recommended configurations for taskwarrior. It will also generate a default configuration file for taskcheck.

Below is an example of taskcheck configuration file:

[time_maps]
# in which hours you will work in each day (in 24h format, if you use e.g. 25.67 you will likely 
# get an error)
[time_maps.work]
monday = [[9, 12.30], [14, 17]]
tuesday = [[9, 12.30], [14, 17]]
wednesday = [[9, 12.30], [14, 17]]
thursday = [[9, 12.30], [14, 17]]
friday = [[9, 12.30], [14, 17]]

[time_maps.weekend]
saturday = [[9, 12.30], ]
sunday = [[9, 12.30], ]

[scheduler]
algorithm = "parallel" # or "sequential", see above
days_ahead = 1000 # how far go with the schedule (lower values make the computation faster)

[calendars]
# ical calendars can be used to block your time and make the scheduling more precise
[calendars.1]
url = "https://your/url/to/calendar.ics"
expiration = 0.08 # in hours (0.08 hours =~ 5 minutes)
timezone = "Europe/Rome" # if set, force timezone for this calendar; timezone values are TZ identifiers (https://en.wikipedia.org/wiki/List_of_tz_database_time_zones)

[calendars.holidays]
url = "https://www.officeholidays.com/ics-clean/italy/milan"
event_all_day_is_blocking = true
expiration = 720 # in hours (720 hours = 30 days)

Tips and Tricks

• You can exclude a task from being scheduled by removing the time_map or estimated attributes.
• You can see tasks that you can execute now with the task ready report.

CLI Options

-v, --verbose: increase output verbosity
-i, --install: install taskcheck configuration
-r, --report: show tasks planned until a certain time