[Matplotlib-users] Problem with axvline in gridspec with log Y axis

[Scott L. <sl...@sp...>]

I get an error when trying to use axvline in a gridspec subplot when the Y axis is set to log scale in matplotlib 1.3.x (and I think recent 1.2.x versions) from github under python 2.7.3 and OS X 10.8.1.  This worked with matplotlib versions from github earlier this year.  Here is a simple code snippet to reproduce the problem, and the error that is generated.  The vertical line is drawn in the vsw sublot if I comment out the dens.axvline line.  The problem seems to be caused by the log scale in the dens subplot.

Any help is appreciated,
Scott

---------------------------------------------------------------------------
import matplotlib as mpl
import matplotlib.dates
import matplotlib.pyplot as plt
import numpy as np
import datetime as dt

starttime = dt.datetime(2012,9,12,20,15)
stoptime = starttime + dt.timedelta(hours=1)
fig = plt.figure(1,figsize=(480.0/72.0, 620.0/72.0), dpi=72)
gs = mpl.gridspec.GridSpec(4, 1, height_ratios=[1.,1.,1.,.67], top=.95, bottom=0.10, left=0.12, hspace=0.12)
vsw = plt.subplot(gs[0], label="vsw", autoscale_on=True, xlim=[starttime, stoptime])
vsw.plot_date(mpl.dates.drange(starttime, stoptime, dt.timedelta(minutes=10)), np.random.random(6))
dens = plt.subplot(gs[1], label="dens", autoscale_on=True, xlim=[starttime, stoptime], yscale='log')
dens.plot_date(mpl.dates.drange(starttime, stoptime, dt.timedelta(minutes=10)), np.random.random(6))
vsw.axvline(dt.datetime(2012,9,12,20,30))
dens.axvline(dt.datetime(2012,9,12,20,30))
plt.show()

---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
<ipython-input-36-781867aaf919> in <module>()
     15 dens.plot_date(mpl.dates.drange(starttime, stoptime, dt.timedelta(minutes=10)), np.random.random(6))
     16 vsw.axvline(dt.datetime(2012,9,12,20,30))
---> 17 dens.axvline(dt.datetime(2012,9,12,20,30))
     18 plt.show()

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib-1.3.x-py2.7-macosx-10.7-intel.egg/matplotlib/axes.pyc in axvline(self, x, ymin, ymax, **kwargs)
   3571             self.transData, self.transAxes)
   3572         l = mlines.Line2D([x,x], [ymin,ymax] , transform=trans, **kwargs)
-> 3573         self.add_line(l)
   3574         self.autoscale_view(scalex=scalex, scaley=False)
   3575         return l

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib-1.3.x-py2.7-macosx-10.7-intel.egg/matplotlib/axes.pyc in add_line(self, line)
   1504             line.set_clip_path(self.patch)
   1505 
-> 1506         self._update_line_limits(line)
   1507         if not line.get_label():
   1508             line.set_label('_line%d' % len(self.lines))

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib-1.3.x-py2.7-macosx-10.7-intel.egg/matplotlib/axes.pyc in _update_line_limits(self, line)
   1525             # identify the transform to go from line's coordinates
   1526             # to data coordinates
-> 1527             trans_to_data = line_trans - self.transData
   1528 
   1529             # if transData is affine we can use the cached non-affine component

/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/matplotlib-1.3.x-py2.7-macosx-10.7-intel.egg/matplotlib/transforms.pyc in __sub__(self, other)
   1204             return self + other.inverted()
   1205         else:
-> 1206             raise ValueError('It is not possible to compute transA - transB '
   1207                              'since transB cannot be inverted and there is no '
   1208                              'shortcut possible.')

ValueError: It is not possible to compute transA - transB since transB cannot be inverted and there is no shortcut possible.