[Pkg-octave-devel] [lukas.reichlin.lists@gmail.com: Re: Debian's control-2.4.1 and quaternion-2.0.2 (Was Re: [OctDev] image 2.0.0 released)]
Lukas Reichlin, upstream author of control and quaternion, asked me why
the last versions of these packages (2.4.1 and 2.0.2, respectively) are
not in testing. I explained to him the mechanism of the Debian freeze
and told him that exceptions for new upstream releases can be made if
there are RC bugs against the version in testing.
I said to Lukas that I could file an RC bug report, provided that he list
the bugs present in the current versions in testing that would make them
unsuitable for release. He then sent the message below to me, which I am
forwarding to your with his autorization.
Please, tell me what you think.
Rafael
----- Forwarded message from Lukas Reichlin <lukas.reichlin.lists@gmail.com> -----
From: Lukas Reichlin <lukas.reichlin.lists@gmail.com>
Subject: Re: Debian's control-2.4.1 and quaternion-2.0.2 (Was Re: [OctDev] image 2.0.0 released)
Date: Wed, 14 Nov 2012 17:58:54 +0100
To: Rafael Laboissiere <rafael@laboissiere.net>
Message-Id: <4CE54E64-C79A-46DA-A44E-92C2B18B1BA5@gmail.com>
On 14.11.2012, at 16:04, Rafael Laboissiere <rafael@laboissiere.net> wrote:
* Lukas Reichlin <lukas.reichlin.lists@gmail.com> [2012-11-14 13:32]:
control news: http://sourceforge.net/p/octave/code/11425/tree/trunk/octave-forge/main/control/NEWS
** feedback
Fixed an argument check which caused false positive error messages.
It was a copy-paste mistake affecting non-square systems.
(Thanks to Tony Olivo)
** sensitivity
Fixed a problem where an error was raised about an undefined function
"issiso".
** impulse, initial, lsim, step
These time response functions now error out correctly when the user tries to
plot multiple systems in one window with mismatching numbers of inputs and
outputs. The problem was caused by Octave's built-in function "size_equal"
which gave false positives for LTI objects. The problem is solved by
overloading the built-in function with a new LTI method "size_equal".
quaternion news: http://sourceforge.net/p/octave/code/11425/tree/trunk/octave-forge/main/quaternion/NEWS
** eq
-- Fix typo bug which gave false positives for the k-component.
-- Avoid error if arguments are non-conformant.
This means that q1 == q2 could give false results. I feel a bit sheepish about that :-)
Note that in "stable", control-1.0.11 does not work properly with octave 3.6.2
Ok, thanks. Let me ask again the crucial question: which are the bugs
in control-2.3.52 and quaternion-2.0.0 that would make those versions
unsuitable for release and that are fixed in control-2.4.1 and
quaternion-2.0.2? Please, give me an answer.
Cheers,
Rafael
Besides the fixes listed above, there are various enhancements which are
a big plus for the users, e.g. control supports system identification,
multisystem plots and there are cleanup efforts under the hood.
quaternion supports the inversion of quaternion matrices.
Best regards,
Lukas
----- End forwarded message -----
Reply to: