On 12/22/2011 04:13 PM, Søren Hauberg wrote: > tor, 22 12 2011 kl. 09:47 -0500, skrev Jordi Gutiérrez Hermoso: >> 2011/12/22 Søren Hauberg <so...@hauberg.org>: >>> tor, 22 12 2011 kl. 09:24 -0500, skrev Jordi Gutiérrez Hermoso: >>>> On 22 December 2011 07:33, Arno Onken <asn...@asnelt.org> wrote: >>>> >>>>> Should kmeans be backward compatible with 3.2.4? >>>> >>>> I don't think it should be. And Søren's fix to replace [~, foo] = >>>> bar() with [tmp, foo] = bar() isn't quite equivalent. Although I don't >>>> think min does this, a function can know if an output parameter was >>>> requested or not and avoid computing it if it wasn't. >>> >>> I really don't think this type of performance tweaks makes much (if any) >>> difference in real life. >> >> No, not in this case, agreed, but I just think that in general >> matching Octave-Forge development to the abilities of some packaged >> version of Octave isn't the best idea. > > I agree with that statement, but (to me) the keyword here is "packaged". > Had you instead said "released", or something similar, I would have > disagreed. I do not see the point in trying to explicitly prevent users > of old Octave versions from using new Octave-Forge packages. > >>From a Debian (or any other distribution) point of view, everything you > say makes perfect sense. I just wouldn't be surprised if some users > (Windows users come to mind) still use 3.2.x, but would like to use more > recent packages. Since it doesn't seem to hurt us to allow them to do > so, I guess I just don't see the point in explicitly ensuring that a > function does not work with 3.2.x. > > Of course, if the change had any impact on performance or similar, I > would feel differently.
I fully agree with that. If there had been any significant impact on performance or on comprehensibility, then I would not have endorsed the change in kmeans. But as it stands, the tiny change moved kmeans from the "only usable on 3.4+" category to the "usable on most system that are currently deployed" category. Arno ------------------------------------------------------------------------------ Write once. Port to many. Get the SDK and tools to simplify cross-platform app development. Create new or port existing apps to sell to consumers worldwide. Explore the Intel AppUpSM program developer opportunity. appdeveloper.intel.com/join http://p.sf.net/sfu/intel-appdev _______________________________________________ Octave-dev mailing list Octave-dev@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/octave-dev