I’m far too depenent on Vim to use a real IDE for development, but I’ve recently
been frustrated by how primitive my python workflow (consisting mostly of
print debugging) is. To fix this I’ve started %running things from inside of
IPython; this makes it easy to open up a pdb and prod the relevant variables
as necessary.
One annoyance remains: when it comes time to fix the error, I want to do it from within the Vim window I already have open. But it would be nice if I could jump straight to the location of the error (known to IPython) rather than having to squint at the stack trace and then perform the necessary navigation myself.
vim-ipython provides various means of passing information back and forth
between Vim and an IPython console, some useful (docstrings) and some not, but
doesn’t support integrated debugging of the kind I want. At least, it didn’t
until this afternoon. Here’s a version that
does.
Various small adjustments need to be made in the message-passing between the Vim plugin and IPython kernel to let Vim read the output of executed commands. (I’m still confused about why the original version of the plugin doesn’t do this.) Then, the following evil little python-Vimscript franken-function jumps to the right place.
function! IPythonTraceback()
python << EOF
command = """
import sys
import traceback
path, line = traceback.extract_tb(sys.last_traceback)\[-1\]\[0:2\]
'edit +%d %s' % (line, path)
"""
reply = get_output(command)
vim.command(reply\[1:-1\])
EOF
endfunctionWhat just happened? A Vim function invokes the python interpreter. The interpreted python program constructs another python program on the fly, which in turn constructs a Vim command. The python program is shipped off to IPython, which runs it and returns the filled-in Vim command, which Vim then executes.
There’s almost certainly a better way to do this, but I fear that if I spend any longer on this the my lifetime gains from the automatic navigation will be overwhelmed by the time wasted on plugin-writing.