Changelog for python-networkx-1.9.1-5.4.noarch.rpm :
Sat Jul 25 14:00:00 2015 seife+obsAATTb1-systems.com
- fix rhel build by conditionalizing \"Recommends:\" tags
- do not hardcode /usr/share/doc/packages but use %_docdir

Wed Apr 29 14:00:00 2015 tbechtoldAATTsuse.com
- Don\'t BuildRequires python-pygraphviz. It\'s not needed.

Thu Oct 30 13:00:00 2014 tbechtoldAATTsuse.com
- update to version 1.9.1:

* Bugfix release for minor installation and documentation issues
- Don\'t BuildRequire/Recommend matplotlib and scipy on SLE11
and SLE12. Both are not available there.

Fri Oct 24 14:00:00 2014 toddrme2178AATTgmail.com
- Add python-decorator in requires to buildrequires

Mon Sep 15 14:00:00 2014 tbechtoldAATTsuse.com
- update to version 1.9:

* The flow package (networkx.algorithms.flow) is completely rewritten
with backward incompatible changes. It introduces a new interface
to flow algorithms. Existing code that uses the flow package will
not work unmodified with NetworkX 1.9.

* We added two new maximum flow algorithms (preflow_push and
shortest_augmenting_path) and rewrote All maximum flow algorithm
implementations (including the legacy ford_fulkerson) output now
a residual network (i.e., a DiGraph) after computing the maximum
flow. See maximum_flow documentation for the details on the
conventions that NetworkX uses for defining a residual network.

* We removed the old max_flow and min_cut functions. The main entry
points to flow algorithms are now the functions maximum_flow,
maximum_flow_value, minimum_cut and minimum_cut_value, which have
new parameters that control maximum flow computation: flow_func
for specifying the algorithm that will do the actual computation
(it accepts a function as argument that implements a maximum flow
algorithm), cutoff for suggesting a maximum flow value at which the
algorithm stops, value_only for stopping the computation as soon as
we have the value of the flow, and residual that accepts as argument
a residual network to be reused in repeated maximum flow computation.

* All flow algorithms are required to accept arguments for these parameters
but may selectively ignored the inapplicable ones. For instance,
preflow_push algorithm can stop after the preflow phase without computing
a maximum flow if we only need the flow value, but both edmonds_karp and
shortest_augmenting_path always compute a maximum flow to obtain the
low value.

* The new function minimum_cut returns the cut value and a node partition
that defines the minimum cut. The function minimum_cut_value returns
only the value of the cut, which is what the removed min_cut function
used to return before 1.9.

* The functions that implement flow algorithms (i.e., preflow_push,
edmonds_karp, shortest_augmenting_path and ford_fulkerson) are not
imported to the base NetworkX namespace. You have to explicitly import
them from the flow package.

* We also added a capacity-scaling minimum cost flow algorithm: capacity
scaling. It supports MultiDiGraph and disconnected networks.
- Add python-decorator as Requires

Mon Dec 9 13:00:00 2013 toddrme2178AATTgmail.com
- Add optional dependencies as Recommends

Sun Dec 8 13:00:00 2013 p.drouandAATTgmail.com
- Update to version 1.8.1
+ No changelog available

Tue Jan 31 13:00:00 2012 saschpeAATTsuse.de
- Don\'t package INSTALL.txt and other docs twice

Thu Jan 12 13:00:00 2012 saschpeAATTsuse.de
- Spec file cosmetics

Wed Jan 11 13:00:00 2012 toddrme2178AATTgmail.com
- Cleaned up spec file
- Renamed package from python-NetworkX to python-networkx to match the module name

Thu Sep 8 14:00:00 2011 alinm.elenaAATTgmail.com
- initial commit