Luc Maisonobe a écrit : > Luc Maisonobe a écrit : >> Phil Steitz a écrit : >>> Looks like we are - at last - ready to cut 2.0. All that remains is >>> >>> 1) resolve MATH-260 and MATH-261 (which I am +1 as marking fixed, based >>> on Luc's last changes) >> +1 >> >>> 2) push remaining unscheduled issues to 2.1 >> Concerning MATH-268 I would consider closing it with won't fix and >> reopening it later if the user still wants it. This is a difficult and >> non-standard feature. >> >>> 3) generate clirr report >>> 4) enhance the maven-changes-report to create proper release notes >> 5) fix the last few bits in Adams-Bashforth and Adams-Moulton that still >> don't work ... Sorry for that but I'm still unhappy with the accuracy of >> these two integrators. I'm working on it. If somebody want to have a >> look at the implementation, please do. Be aware it is quite difficult. I >> have put in the class javadoc as much as I could to explain the >> underlying theory because I only found sparse documentation on the web >> and had to redevelop some equations myself. > > I got it! It now seems to me the problem was numerical instability, not > an error in the equations or implementation. Step size growth should be > very slow for these integrators, I've changed the initial automatic > setting to a safer value. > > Now there is nothing left on my side that would delay 2.0 anymore. > >>> I will do the RM-ing if no one else wants to but would appreciate help >>> in writing a summary of API changes to add to the release notes. >> I can help on that. I'll first try once again to understand the problem >> with the integrator and if it takes too long I'll give up for a while >> and will put a notice for users that these are experimental >> implementations. Then I'll write a summary on the various API changes I >> know about (mainly ODE, optimization, root solvers and linear algebra). > > Where do you want me to put such a summary ? I think the release notes > are automatically generated from the changes.xml file, so I probably > need to put it somewhere else.
I've set up a first version of the release notes yesterday, checked the remaining bits about serialization and improved tests coverage today (we now have more than 2000 unit tests with a test coverage of 89%). The current state is ready for publication in my humble opinion. I will not have much more time this week. Phil, if you are still OK for managing the release, I'll let you do it when you want. Thanks, Luc > > Luc > >> Luc >> >>> Phil >>> >>> --------------------------------------------------------------------- >>> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >>> For additional commands, e-mail: dev-h...@commons.apache.org >>> >> >> --------------------------------------------------------------------- >> To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org >> For additional commands, e-mail: dev-h...@commons.apache.org >> >> > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org > For additional commands, e-mail: dev-h...@commons.apache.org > > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@commons.apache.org For additional commands, e-mail: dev-h...@commons.apache.org
