When running code across various machines occasionally I will see an error that I would like to debug. In ipython, simply typing "debug" will let me interactively navigate up/down through the stack trace and see what was causing the issue.
The debug only works on the most recent exception. Is it possible to dump something to a file upon exception so that a command like debug can be called later to explore all the elements that were in memory? This is especially useful when code is being run on a parallel environment / cluster and I want to see why a particular instance of the code failed (since it is not always possible to reproduce locally).
与恶龙缠斗过久,自身亦成为恶龙；凝视深渊过久,深渊将回以凝视…
