error bars in log-log plot, and upper/lower limits

Hi all,

I'm fairly new to matplotlib, and so far, I have not found an (easy)
way to plot error bars on a log-log plot. Is there a way to do that
easily (eg through the pylab interface), or do I have set up things
more manually?

Related to that, I'm also looking for a way to neatly produce upper
(or lower) limits on figures, either log-log or normal or a
combination. This is generally a problem in every plotting package,
but perhaps someone has a good pointer to this. Drawing arrows helps
(for example, down-pointing arrows for y-upper limits), but his often
needs to be done separately, as I do now in gnuplot (ie, the arrows
stand just on their own, and are not seen as limits by the plotting
package: for y-upper limits, I need to draw the horizontal error bars
separately, as well as draw the non-limit points separately).

If there's no relatively straightforward way, I'd be inclined to delve
into the pylab.errorbar routine, to try and change that so that's it's
suitable for log-log plots, and accepts limits (by eg supplying None
as an element of the Nx2 numpy array; obviously, limits only work for
asymmetric errorbars). But that may simply not be the best thing to
do, so any pointers to that would also be welcome.