GitLab

Because Java does not handle a native support of UNix domain sockets...

Now get the taktuk id

Can do AND and OR.

* Close a mpi bridge upon destruction
* Cancel timer upon task ending
* add logs

Work needs to be done before declaring wanting to transit between
running and done states. This work is, asking the underlying transport
to kill its instances and clean stuff on the remote computers. Thus,
declaring that the work is done is wrong. The work is done when this
step is done.

To enable this asynchonous possibility a task can now ask the execution
of a callback function when a changing state is possible. It's a way to
execute an out-of-workflow operation still inside the state-machine
transitions.

Providing better event handling and on transitions triggers/dependencies
possibilities.

Because, all the undelying tasks mechnism is buggy and needs a refactor

A cancelled task may not get to the done state ?

Some environnement like Spark do not send the argument with only one
dash. Allow the MPI_EXECUTOR to send arguments with the full name.

Because on broadcast a regex is given in dest field. And the problem
resolved by the previous commit only touched the command field.

Because we are building regex from random strings, we need to escape
them.

When a node die TakTuk may trigger two different messages, internal data
structures need protection to avoid accessing an already handled error.

Isengard still contains bits of code that are not valid, mostly stuff
about variables that were deleted some time ago but are still being
accessed.

Hopefully this is one of the last ones.

To allow the creation of a new connector, we need to extract every
uncesseray files/code from the isengard folder. This way, creating a new
connector will be to provide an other isengard_like folder with only
specif stuff.