[IPython-User] [Ipython.parallel] How to stop on (first) error?

Vincent Dubourg vincent.dubourg@gmail....
Fri Feb 28 03:13:32 CST 2014


Hi list,

Let say I use a load balanced view of my 4 engines client.
Then I use the map method of that view to execute some (randomly buggy)
function of mine in parallel over a list of 600 inputs.
For now it seems it runs the whole list and eventually fails with 98
errors. I'd rather have it failing as soon as it mets the first error.

Is it possible to configure IPython.parallel so that it stops running on
the first error it mets? So that I could then debug that particular error
by looking at the files in the current working directory of the engine that
met the error (the error is caused by a third-party software I run through
subprocess).

I hope this is clear! I could not find an answer to my question on the web,
so I raise it with my words here...

Cheers,
Vincent
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.scipy.org/pipermail/ipython-user/attachments/20140228/c82e3fad/attachment.html 


More information about the IPython-User mailing list