Slic3r 0.9.8 is out

[mhackney]

1/17/2013 - version 0.9.8 has been released.

New features:
acceleration control (advanced setting) to control behavior for perimeters and infill
customizable G-code to be executed on tool change ([previous_extruder] and [next_extruder] placeholders can be used, as well as nested placeholders like [temperature_[next_extruder]])
automatic detection of support material threshold angle (leave it to zero and Slic3r will check your layer height and extrusion width to detect unsupported overhangs)
experimental feature for vibration limiting (not needed in 99.9% of cases, might be handy with some shaky printers)
support for a Sailfish G-code flavor (I can't test this one, so please provide feedback)
Improvements:
plater will show vertical projection (with holes) of objects instead of their convex hull (limited to models with less than 2000 triangles for now - performance reasons)
gap fill is always disabled when fill density is zero; you can also set gap fill speed to zero to disable it
slight memory and time savings
automatic extrusion width defaults to way saner behavior when using low layer heights, avoiding infill to be too sparse
Changes:
cooling is enabled by default now
Bugfixes:
many (all?) spikes/blobs removed, causing not very smooth surfaces when high-res models were processed
wrong direction was detected for many bridges
infill every 'n' layers was stalling or generating tiny useless extrusions
the experimental inwards move that was recently introduced to avoid retracting on external perimeters was sometimes moving outwards (!) causing a little blob
some top/bottom solid shels were not treated as such
some holes were filled/bridged incorrectly
crash on Windows when username or STL/G-code paths contained non-ASCII characters
interface support material layers were clipped badly causing extrusions running into object perimeters
slow infill patterns were not working anymore
extra perimeters were not always applied when needed
custom layer-change G-code was applied before Z change
fixed regression causing inefficient infill paths
restart_extra_length didn't work correcty with negative values
retraction lift was not playing well with retract_extra_length in multi-extruder environments