- Bug: In AgdsQuery::goCallBack(), take account the instance transformation.
This problem was not showing until now because there wasn't any instances...
- New: Path module to provide a ligthweight alternative to boost::filesystem.
The compatibility change in boost was giving more and more touble and was
not worth it. Path is aimed to provides the same services, but with a
better portability.
- New: In PyDataBase, added method for getDB (the correct name, as alias)
and getRootLibrary().
- New: In PyComponent, added support for the optional BasicLayer argument
of getBoundingBox().
- New: In PyEntity, added support for conversion to/from RoutingPad.
- New: In PyHyperNet, added support for getNetOccurrences().
- New: In PyLibrary, added support for getLibrary().
- New: In NetExternalComponents, added support for setExternal().
- Bug: In CMakeLists.txt, when our custom macro "setup_boost()" is used,
do not call find_package() afterward, it corrupt the library pathes
from '/usr/lib/' to '/usr/lib/lib/'.
- New: In hurricane/doc, added documentation RoutingPad, HyperNet & Query.
- New: In isobar, added support for RoutingPad.
- Change: In PyComponent, getCenter() method moved to it's right place in
the base class.
- Change: In PyPath, added duplicated methods (ApplyOn <-> applyOn and
Invert <-> invert).
- New: In MaukaEngine::Construct(), do not try to initialize the slice height
from a model with an empty AbutmentBox. Instead, signal an error (the model
is likely to be missing, it has been initialized from the COMPONENT).
- Change: In MaukaEngine::Construct(), clearer error message when there is duplicated
instances. Show the full occurrence path of each conflicting instance.
- Change: In PyHurricane, add type object for all subtype of Component, then
we can use the Python "isinstance()" function to know the true type of a
component. Useful when looping over a collection of Component.
Name clash: We have used the class name as the name of the constructor
inside the Hurricane module, at a time we where less acquainted with
Python, so the real class are named 'PyVertical' for example, while
'Hurricane' is the constructor function. Should correct this in the
future but it's a huge work as all the Python scripts must be editeds...
- Change: In PyNet, adds the "getComponents()" Collection accessor.
- New: In Occurrence, added method getCompactString() to display a more
compact version of the occurrence.
- New: PyQuery & PyQueryMask expose the Query classe. Only partially implemented,
and with the start of the template based interface disabled.
- New: Complete rewrite of builCoriolis.py & merge with easyChams to provide
a graphical interface. Asides to the GUI, when run in graphic mode all
settings are kept from run to run. Meaning that once the initial setup
is done, a user may completly forget where things are... The new tool
is named <ccb> for 'Coriolis & Chams Builder'.
The Builder is now a module, that is split into multiples files.
The old buildCoriolis.py is kept for the time beeing to ensure a smooth
transition.
- Change: In CgtMain, rename the namespace boost::program_options from boptions
to bopts (to be identical to KiteMain).
- Change: In CMakeLists.txt, the C++ binary is renamed from 'cgt-cpp' toward
cgt.bin, again to follow the same naming scheme as Kite.
- Bug: In PyHurricane, PyTypeViaLayer was *not* reference incremented, causing
the type to be desallocated when the Python interpreter was finalizing.
This was causing <isis> to fail.
- New: In buildCoriolis.py, in the svnCheckout command, if the source
directory do not exists, create it (recursively).
- Change: In buildCoriolis.py, in the svnCheckout command, take the svn
tag version number in account now...
- Bug: In Cs1Candidate::consolidate(), the loop on the conflicting segment was
going one step to far. "i" must be bound to _conflicts.size()-1.
- Change: kite-text is renamed into kite.bin. It features the router only
with a subset of options consistent with cgt (python). Mainly used for
debugging purposes (gdb, valgrind, ...).
- Bug: In Histogram::addSample(), reject the sample if iset is superior
OR EQUAL to the sets size...
- Change: In etc/mauka.conf, sets "mauka.standardAnnealing" to True. Otherwise,
when hMetis (4-part) is not used, this leads toward very poor placements.
- New: In UsersGuide, add a third style variant for the soc-extras site.
Jean-Paul Chaput
Roselyne Chotin
Ramy Iskander