Run shell commands within Unixqueues
This module is not thread-safe. See the module Shell_sys
for
more information.
Signal handlers: It is important to have a number of signal handlers installed for proper function of the engines. It is recommended to call the pair of functions Shell_sys.configure_job_handlers and Shell_sys.install_job_handlers for this purpose. This is not done automatically.
Note that this has a global side effect on the whole process, because there is only one set of signal handlers.
This type of engine also returns the job
and the job_instance
.
This engine corresponds to Shell.call. The command list is
executed until the job is done. In this case,
the status of the engine changes to `Done
or `Error
.
If not ignore_error_code
, failed jobs are reported by the
error status `Error Subprocess_failure
. If ignore_error_code
,
failed jobs are ignored, and result in the status `Done
(however, jobs terminated by signals are still reported as
errors).
For the other arguments see Shell.call.
When the engine goes to `Error
or `Aborted
, the job
is terminated ( Shell_sys.abandon_job ).
All examples presented for the [root:Shell] module can be easily rewritten
for Shell_uq
. For example, to call "ls" and collect the result in a
buffer, use this piece of code:
let ues = Unixqueue.create_unix_event_system();;
let b = Buffer.create 10;;
let e = new call_engine ~stdout:(Shell.to_buffer b) [ Shell.command "ls" ];;
Unixqueue.run ues;;
let r = Buffer.contents b;;
This means, one simply creates a call_engine
instead of invoking
Shell.call, and then runs the queue. Note that you must look at
e#state
to find out whether the engine e
produced an error, no
exception is raised in this case.
It is allowed (and somehow the purpose of this module) to create several job engines, and let them run in parallel.