[general need] Mia loses the name of the brick in case of a calculation launch error (e.g. a pipeline without input parameters)

[servoz]

The minimum steps to reproduce :

• Open Mia
• Open a project
• In the pipeline manager, drag and drop the SPM Smooth brick from mia_processes, into the editor
• Click on the "Run pipeline" button (this is an error as we have not yet defined the input parameters, but errors should be handled in Mia)
• A pop-up window opens indicating that a mandatory parameter for the pipeline is missing (note that the Smooth brick/process is called main, which is not appropriate) => Except for the problem of the name of the brick, the case is well managed now (following the previous opened ticket on this subject).

If we now want to get things right:

• Right click in the brick and choose "Export all unconnected records" (all nodes turn blue, why ?)
• Select a in_files parameter using the Filter button, in the controller
• Click on the "Run pipeline" button
• Calculation runs correctly, but:

We see on the status bar: Pipeline "NoName" has been correctly run We see on the stdout: And we see in DataBrowser for the Bricks tag (this is the most important problem, as we lose the data history in this case) :

EDIT: If we hit the main button, mia crashs with this exception:

Traceback (most recent call last):
  File "/casa/home/Git_projects/populse_mia/python/populse_mia/user_interface/data_browser/data_browser.py", line 2177, in show_brick_history
    self.data_browser.main_window)
  File "/casa/home/Git_projects/populse_mia/python/populse_mia/user_interface/pop_ups.py", line 5034, in __init__
    self.node_selected(full_brick_name[0], pipeline.nodes[full_brick_name[0]])
KeyError: 'main'

[servoz]

Other name issue has been noticed. See the ticket #263.

[servoz]

The "main" name comes from the capsul.pipeline.pipeline_workflow.workflow_from_pipeline() method when the pipeline is in fact a single process.

Maybe I'm missing something (maybe there's a reason for using 'main' as the name in all cases?) but as this is a pipeline with a single process, the node must take the name of the class + '_1'.

So, I think this line should be: new_pipeline.add_process(pipeline.name.lower() + "_1", pipeline)

@denisri, what do you think? If you don't see a side effect I'm missing, I can make the change directly in capsul.

[denisri]

No problem, it's fine... it was just a bit of code done too quickly.

[servoz]

Done.