octave-forge-control-3.3.1-bp154.1.34

- Update to version 3.3.1:
  * Fixed time response for first order systems.
- Changes from version 3.3.0:
  * New function rlocusx providing gain, damping and frequency for
    selected poles together with the possibility to generate open
    loop bode plots and simulations of the closed loop for
    selected closed loop poles.
  * Fixed several issues in bode, c2d, lsim and step.
  * Fixed issues in legends of time and frequency responses when
  * requested for multiple systems.
  * Fixed usage of deprecated LAPACK routines.
  * Added discretization method 'foh' to c2d.
  * Added tests for c2d with 'foh' and to dlyapchol.
  * Fixed transposing a tfploy.
- Drop control-gcc-errors.patch: No longer required as fixed
  upstream.

- Update to version 3.2.0
  * new function sgrid
  * New function: ss2ss
  * Fixed many warnings
  * added demos to rlocus, pzmap, bode, nichols, nyquist, impulse, lsim,
    ramp and step
  * improved pzmap to plot with only zeros
- Update to version 3.1.0
  * New function: damp
  * Changed nelem to numel and length to numel to remove the warnings
    during installation.
  * The Impulse section has been rewritten, and a new file - imp_invar.m -
    added. The old system tried to use the "zoh" to get an impulse response
    but that is impossible.
  * The new function "imp_invar.m" converts a Laplace tranfer function
    to a discrete tranfer function.
  * New functions: acker, dsort, esort.
  * Function lsim now plots inputs as well.
  * Class 'ss' should now work with complex inputs.
  * Extend nyquist plots.
  * Fixed warnings with newer Octave versions.

- Update to version 3.0.0
  * Substantial parts of the most vital internals have been rewritten
    and streamlined in order to improve compatibility and reliability.
  * LTI models:
    Improved handling of key-value pairs in all constructors for LTI models.
    Support inheritance of LTI properties in many variations.
    Accept [] as an alias for unspecified sampling time (Ts = -1).
    Automatic vector expansion for input and output names.
  * dss
    Fixed a long-standing but undiscovered bug where 'dss (sys)'
    overwrites sys.e with an identity matrix even if sys.e was not empty.
  * ss
    Support for the special cases has been added.
  * tf, zpk, filt
    Restrict transfer function variable to the candidates 's', 'p' for
    continuous-time and 'z', 'q', 'z^-1', 'q^-1' for discrete-time models.
  * bode, bodemag, nichols, nyquist, sigma
    It is now possible to specify a frequency vector of length 1.
  * step, impulse, initial, ramp
    The time response functions now raise an error or warning if invalid
    arguments are passed to them.  As it was the case for frequency response
    functions, they used to simply ignore invalid arguments.
  * lsim
    Support logical arrays as input signal.  Before, logical arrays were
    simply ignored without notice, which could lead to unexpected results.
    Additionally, lsim received the same improvements as the other time and
    frequency response functions regarding invalid arguments.
  * pzmap
    Improved argument checks.
  * augstate
    New function for state-space models.  Adds the state vector to the output
    vector.
  * connect
    The function no longer raises an error for name-based interconnections
    if some systems have common input names and one of these inputs are to
    be kept.
  * estim
    Add input/output/state names to the returned system.  Support name-based
    selection of inputs (arguments 'sensors' and 'known').
  * * sumblk
    Add signal names as input/output groups if argument  n > 1.
  * * minreal
    If state-space models already have a minimal realization, then no state
    transformations are applied to the returned model.

- Update to versiob 2.8.5
  * Bug fixes
  * Matlab compatibility improvements

- Update to version 2.8.4
  * Bug fixes and minor changes

- Update to version 2.8.3
  * Bug fixes and minor changes

- Update to version 2.8.2
  * Bugfix release

- Update to version 2.8.1
  * tf
  - Revamped the modifications of TB01ZD from control version 2.8.0.
    The function now uses LAPACK routines to find the maximum norm element.
  - Improved error handling when converting state-space models.

- Update to version 2.8.0
  * zero
    Compatibility fix in the C++ oct-file to support the upcoming Octave 4.0.
  * isminimumphase
    Reverted changes from control-2.6.6.  According to the
    definition of Byrnes/Isidori, the function tests whether the system
    has asymptotically stable zero dynamics. The poles are not tested.
  * series
    Fixed a bug which gave unnecessary error messages when the
    function is called with 4 arguments.
  * tf
    Resolved issues in MIMO state-space to transfer function
    conversion by implementing a more robust algorithm in SLICOT
    routine TB01ZD.
  * Support for 64bit indexing option.

- Split from octave-forge package, version 2.6.6
- Fix GCC errors: no return in non-void function
  * control-gcc-errors.patch