WIP: Enh/provenance #451
WIP: Enh/provenance #451
Conversation
|
Could you update the docs with a page explaining this feature? |
|
I'm happy to merge it, but we need some form of documentation. |
| from nipype.pipeline import write_prov | ||
| from nipype.external.provgraph import prov_to_file | ||
| g = write_prov(eg, 'workflow.json') | ||
| prov_to_file(g, 'workflow.pdf', use_labels=True, format='pdf') |
There was a problem hiding this comment.
You needto use a different example for this. I was trying to generete this graph, but it has been running for over 24h now...
…pype execution graphs.
* better artifact handling * better capture of runtime info
| @@ -1117,6 +1336,31 @@ def _process(drain=0): | |||
| return runtime | |||
|
|
|||
|
|
|||
| def get_dependencies(name, environ): | |||
There was a problem hiding this comment.
@yarikoptic and @Hanke : could you suggest some better ways of getting at this info without hooking into extra dependencies? we can definitely improve this later, but this branch is otherwise in reasonable shape to merge.
There was a problem hiding this comment.
but what do you mean by "this info"? ;)
if only a list of linked to libraries -- I guess, it could not be much better. If you actually want to "look inside" and figure out what other libraries/plugins this beast might be loading/opening (directly) then probably investigation of symbols provided elsewhere and associating them with one you can find used in this binary (like I believe https://github.com/rocksclusters/FingerPrint/blob/HEAD/FingerPrint/plugins/find-requires does) -- then there is more TODO.
If you actually want to get all dependencies needed during run-time, I guess you would need first to run the command via e.g. strace...
There was a problem hiding this comment.
BTW hello @satra from Boston children museum's au bon pain (kids are running, I am working ;) )
|
I'm happy to merge it, but we still need some (at least rudimentary) documentation. |
|
@chrisfilo what kind of documentation would you like? in most cases this is still behind the scenes. we still need to build up apps on top of it (e.g. reporting). i could point to the nidm notebook somewhere. would that suffice for now? and that reminds me that i do need to add a nipype vocabulary guide. but i'll do that in a different pr. |
|
@chrisfilo can i merge this? |
|
Go for it. Sorry for delay. On Sun, Sep 8, 2013 at 8:01 PM, Satrajit Ghosh notifications@github.comwrote:
|
Preliminary PROV support in nipype