I can imagine, during development, that we need to test an integration of task and don't want to actually wait until the scientific processing of the tool has finished. Especially when we're doing multiple rounds of adjustments of special modules and the exact content of the output is not in question.
So I suggest that admins and developers be provided with a special panel at the bottom of all Boutiques tools, generated by the integrator, that would have an option named something like "generate fake outputs". The panel would list all the outputs entries defined in the descriptor, and let the user either provides a dummy name for a file or folder, or provide the ID of any CBRAIN file that woudl be used as a replacement output.
E.g. the BoutiquesDemo tool that comes with CBRAIN has one output defined, as can be seen in
and the corresponding panel would look like:
+-------------------------------------------------+
| Generate fake output files? [X] |
| ID 'du_report_out' ("Output report") [________] |
+-------------------------------------------------+
So the developer can enter in the input box [______] something like
"fakefile" : cbrain generates a dummy file with a few bytes in it
"fakedir" : cbrain generates an arbitrary dummy folder
"123456" : cbrain installs the userfile 123456 in the workdir where the output should appear.
This mechanism could also be used just to test that a remote production installation is working end-to-end except for the actual computation part.
I can imagine, during development, that we need to test an integration of task and don't want to actually wait until the scientific processing of the tool has finished. Especially when we're doing multiple rounds of adjustments of special modules and the exact content of the output is not in question.
So I suggest that admins and developers be provided with a special panel at the bottom of all Boutiques tools, generated by the integrator, that would have an option named something like "generate fake outputs". The panel would list all the outputs entries defined in the descriptor, and let the user either provides a dummy name for a file or folder, or provide the ID of any CBRAIN file that woudl be used as a replacement output.
E.g. the BoutiquesDemo tool that comes with CBRAIN has one output defined, as can be seen in
cbrain/BrainPortal/cbrain_plugins/cbrain-plugins-base/boutiques_descriptors/boutiques_demo.json
Line 57 in 02d9156
and the corresponding panel would look like:
So the developer can enter in the input box [______] something like
"fakefile" : cbrain generates a dummy file with a few bytes in it
"fakedir" : cbrain generates an arbitrary dummy folder
"123456" : cbrain installs the userfile 123456 in the workdir where the output should appear.
This mechanism could also be used just to test that a remote production installation is working end-to-end except for the actual computation part.