Hi Martin, Points below, well taken and I am +1 to move forward.
Cheers, Chris On Jul 18, 2012, at 11:03 AM, Martin Desruisseaux wrote: > Hello Chris > > Thanks a lot for you email, I'm glad to see there is interest! > > > Le 18/07/12 16:46, Mattmann, Chris A (388J) a écrit : >> Ah, NetCDF, I knew that you and I had crossed paths before too! :) Maven, >> NetCDF Java support, anyone? :) > Yes, I still in touch with some UCAR and Readings developers for various > aspects related to NetCDF. > >> Yep, that's how Apache works. We have CI systems (ci.apache.org), we leverage >> unit tests, we generally care about the code, but more importantly, we >> *care* about >> the people here and are all mutually friendly to one another. > Actually I had a few years of experience with an Open Source community in the > past. At that time, it was on a centralized versioning system (SubVersion). > The project had a friendly start, but a few years later some tensions > happened. My analysis is that they were conflicts in the goals and > constraints of different developers: > > * Conflicts in the goals between the "scientist inclined" and the > "mass-market inclined", since the former want computation results they can > trust while the later want pretty pictures to show fast. While both needs are > perfectly valids, my (maybe egocentric) wish would be to give priority to the > former, because I think it is easier for a "mass-market inclined" to take a > trusted library and makes it more flexible for his needs, than for a > "scientist inclined" to make an extensive review of a library for the places > where correctness may be compromised. (Note: this discussion ignores the > "security inclined" because I have no experience in that area, but it would > probably be an other factor to consider.) > > * Conflicts in the constraints between developers working for different > institutions or companies, since a company may have strong deadlines for > their releases. It may result in some peoples pushing strongly for a work to > be injected in the code base, while some others feel uncomfortable but lack > of time for alternative proposals. For this particular issue, I put my hopes > in Distributed Versioning Systems. I'm a supporter of Joshua Bloch statement > "In case of doubt, leave it out": I would like the SIS project to move > relatively slowly, to commit only things we feel confident about, and let > each institutions/companies manage their own DVS clones with all the > branding-new code they wish. > > * Conflicts in the choice of technologies (logging framework, etc.). While I > think everyone agree for letting the choice to users, I though (maybe > naively) that the most consensual technology would be the one bundled in the > standard JDK, when applicable. It is often possible to use the JDK API in a > way that still allow freedom of choice to users. But I realized that this > approach still sometime controversial. > > > So based on the above, I wonder what peoples feel about the following? > > * Focus on a "trusted" library, eventually with branches for peoples > who need trimmed-down versions; > * Relatively "slow" commitment, with institutions/companies managing > their own clones if needed > > > Martin > ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Chris Mattmann, Ph.D. Senior Computer Scientist NASA Jet Propulsion Laboratory Pasadena, CA 91109 USA Office: 171-266B, Mailstop: 171-246 Email: chris.a.mattm...@nasa.gov WWW: http://sunset.usc.edu/~mattmann/ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ Adjunct Assistant Professor, Computer Science Department University of Southern California, Los Angeles, CA 90089 USA ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
