task-sync(5) User Manuals task-sync(5)NAMEtask-sync - A tutorial for the task(1) data synchronization capabili‐
ties.
DESCRIPTION
Taskwarrior has built-in support for synchronization, which can be used
to keep two task databases up to date, regardless of which one is used.
This capability can also be used to keep a backup copy of your task
database on another machine.
Taskwarrior can use various protocols for transferring the data.
HOW IT WORKS
If you were to manually attempt to keep two separate task databases up
to date, you would need to inspect both databases, and detect changes
that occurred in each one. Those changes would need to be migrated to
the other database, while being careful not to miss a change, and not
to confuse an 'add' in one with a 'delete' in the other.
The synchronization feature does just this. It can transfer task data‐
bases, compare tasks, and apply changes where necessary.
NEW COMMANDS
Taskwarrior has 'pull', 'push' and 'merge' commands which perform the
steps necessary to move files around and combine them. In the common
use case, you would only need to use the 'merge' command. These com‐
mands take an argument that is a URI, which indicates where the remote
database resides.
To be clear, the local database always refers to your ~/.task directory
(unless overridden), and the remote database is always specified by
URI.
MERGE
The merge command will fetch task data via URI and combine it with the
local task database. The syntax is:
task merge [<URI>]
The URI is optional if the merge.default.uri configuration variable is
set. The URI may point to a different directory, or it may be a dif‐
ferent computer. Here is an example of the merge command:
$ task merge ~/work/
This URI (~/work/) is a path name, which means the remote database is
on the same computer. Taskwarrior will fetch the data from the URI,
and merge it with your local data in ~/.task.
When complete, you will be asked whether you would like to push the
combined data back to the remote location specified by the URI. This
is useful if you are keeping two task databases synchronized, but it
can be turned off. See CONFIGURATION.
Note that a merge operation is not atomically reversible. You could
however run the 'task undo' command repeatedly to undo the effects.
PUSH
The push command will copy the local task database to the specified
URI. The syntax is:
task push [<URI>]
The URI is optional if the push.default.uri configuration variable is
set. This command is useful for making backup copies of your task
database.
Note that the task files at the location specified by the URI are sim‐
ply overwritten, so don't expect any merging to occur. Misused, push
can be dangerous.
PULL
The pull command will copy a task database from a URI to the local task
database (~/.task by default). The syntax is:
task pull [<URI>]
The URI is optional if the pull.default.uri configuration variable is
set. This command is useful for restoring a backup copy of your task
database.
Note that your local task database files will be simply overwritten by
the files obtained from the location specified by the URI, so don't
expect any merging to occur. Misused, pull can be dangerous.
URI TYPES
The most basic URI is a path name on the local machine. An example
would be:
/home/bob/.task/
All the other URIs allow access to remote machines. The first uses SSH
and scp (either form can be used):
ssh://[user@]host[:port]/absolute/path/to/.task/
[user@]host:/absolute/path/to/.task/
In both cases paths are considered to be absolute. You can specify
paths relative to the users home directory as follows:
ssh://[user@]host[:port]/.task/
[user@]host:.task/
or even shorter
[user@]host:.task/
Remark: Since taskwarrior simply calls the scp binary you can specify
very much anything that scp would accept, e.g. host configurations from
~/.ssh/config or ~username expansion:
ssh://configured-host/~[username]/.task/
configured-host:~[username]/.task/
Rsync is another supported protocol that minimizes network traffic, by
a clever algorithm that doesn't copy files that have not changed:
rsync://[user@]host.xz[:port]/path/to/.task/
Curl supports several protocols that can transfer data using HTTP,
HTTPS and FTP:
http://host[:port]/path/to/.task/
https://host[:port]/path/to/.task/
ftp://[user@]host[:port]/path/to/.task/
You can use single quotes to encapsulate user names that contain delim‐
iting characters like '@', '/' or ':', e.g.:
ssh://'user@name'@host/
Remember to escape the quotes on your shell:
$ task push ftp://´user@name´:host/
CONFLICTS
When modifications on the local and remote machine conflict, for exam‐
ple if both machines change the project name of the same task to dif‐
ferent values, then Taskwarrior automatically selects the most recent
change. Thus, there are no conflicts.
EXAMPLE - Backup on another machine
One very good use of 'push' is to make backup copies of your task data‐
base in another location. Suppose your task database is kept in the
usual place, in the ~/.task directory, and you wanted to make a backup
copy in ~/backup. You would use this command:
$ task push ~/backup/
This would copy the files in ~/.task to ~/backup, overwriting the files
that were already in ~/backup. To backup your files to another
machine, you could use:
$ task push user@host:backup
This could be improved by setting the push.default.uri configuration
variable and then relying on the default, like this:
$ task config push.default.uri user@host:backup
and then you need only run the push command:
$ task push
and the default push URI will be used. If you wanted to restore a
backup, you simply use the pull command instead:
$ task pull user@host:backup
This can be simplified by setting the pull.default.uri configuration
variable and then relying on the default, like this:
$ task config pull.default.uri user@host:backup
Note that pull and push will blindly overwrite the task files without
any merging. Be careful.
EXAMPLE - Keeping two task databases synchronized
The most common synchronization will be to keep two task databases syn‐
chronized on different machines. Here is a full example, including
setup that illustrates this.
Suppose there are two machines, named 'local' and 'remote', for sim‐
plicity. Taskwarrior is installed on both machines. The different
machines are indicated here by the prompt. Suppose Alice enters two
tasks on her local machine:
local> task add Deliver the new budget proposal due:tuesday
local> task add Set up a meeting with Bob
Then later adds a task on the remote machine:
remote> task add Present the budget proposal at the big meeting
due:thursday
Now on the local machine, Alice merges the two task databases:
local> task merge alice@remote:.task
Would you like to push the changes to 'alice@remote:.task'? Y
Taskwarrior has combined the two task databases on local, then pushed
the changes back to remote. Now suppose Alice changes the due date for
task 1 on remote:
remote> task 1 due:wednesday
Now on the local machine, Alice sets up a default URI and autopush:
local> task config merge.default.uri alice@remote:.task
local> task config merge.autopush yes
Now Alice can simply run merge to make sure that the new due date is
copied to the local machine:
local> task merge
This time the URI is determined automatically, and after the merge the
files are pushed back to the remote machine. In a similar way, the
remote machine can also be configured to merge from the local machine
and push back to it. Then it is just a matter of Alice remembering to
merge now and then, from either machine, to have her data in two (or
even more) places.
CONFIGURATION
By setting these configuration variables, it is possible to simplify
the synchronization commands, by relying on the defaults or alias
names.
merge.autopush=yes|no|ask
This controls whether the automatic push after a merge is per‐
formed, not performed, or whether the user is asked every time.
The default value is 'ask'.
merge.default.uri=<uri>
Sets a default URI so that just the 'task merge' command be run
without the need to retype the URI every time. You can also use
this configuration scheme to set alias names, e.g. set
merge.desktop.uri and run 'task merge desktop'.
push.default.uri=<uri>
Sets a default URI so that just the 'task push' command be run
without the need to retype the URI every time. You can also use
this configuration scheme to set alias names, e.g. set
push.desktop.uri and run 'task push desktop'.
pull.default.uri=<uri>
Sets a default URI so that just the 'task pull' command be run
without the need to retype the URI every time. You can also use
this configuration scheme to set alias names, e.g. set
pull.desktop.uri and run 'task pull desktop'.
Note that, when using SSH/scp, hostnames will be expanded due to the
ssh configuration file ~/.ssh/config.
EXTERNAL DEPENDENCIES
Depending on the URI protocols used, the utilities 'scp', 'rsync' and
'curl' must be installed and accessible via the $PATH environment vari‐
able.
If you have deleted your ~/.task/undo.data file to save space, you will
be unable to merge. The change transactions stored in the undo.data
file are used for synchronization.
CREDITS & COPYRIGHTS
Copyright (C) 2006 - 2013 P. Beckingham, F. Hernandez.
The sync capabilities were written by J. Schlatow. Parts copyright (C)
2010 - 2013 J. Schlatow.
Taskwarrior is distributed under the MIT license. See http://www.open‐
source.org/licenses/mit-license.php for more information.
SEE ALSOtask(1), taskrc(5), task-faq(5), task-color(5), task-tutorial(5),
ssh_config(5)
For more information regarding task, the following may be referenced:
The official site at
<http://taskwarrior.org>
The official code repository at
<git://tasktools.org/task.git/>
You can contact the project by writing an email to
<support@taskwarrior.org>
REPORTING BUGS
Bugs in task may be reported to the issue-tracker at
<http://taskwarrior.org>
task 2.2.0 2013-04-07 task-sync(5)