I found that v0.9.4 generated a file 19 MB in size compared to SF's 1.6 MB. The PC running the printer was trying to pass that many gcode lines to the printer that the printer kept slowing down and speeding up. I was trying to print what was basically a tube. When printing 2 hollow boxes (thin wall test) it joined them together with 1 thicker wall.
I have also found that all my thin wall tests fail in that they have thinner walls than expected (I set the thickness manually) unless the extrusion multplier is set to 1.5 or so but printing anyother object with the extrusion multiplier set like this results in grossy to much material being extruded. This susgests that my E values are in fact correct (they have been checked a number of times after making changes - such as replacing/modifying the hobbed bolt). It's almost like the extruder needs to run at or higher than some minimum speed or it starts to slow down/stall - how you would expect a DC motor to behave.