Hi Philippe, I totally agree with all your comments and this is exactly the position of the ROOT team. Cheers, Fons. > > I think that the current cvs repository plans from the ROOT team > are an interesting step towards improving on the current open and > collaborative environment but it should go further in a not too distant > future. > > The cvs repository that Fons is going to introduce does have some nice > advantages over the current situation. One immediate advantage is > that from now on, any user will be able to easily access the source > code for any particular release, and not just the latest one as > in the current scheme. Another possible advantage, even-though > this is obviously not going to be common, is to make it easier > to introduce bug fixes for older versions. This might be necessary > for some experiments that are soon going to need to select a production > release from which they will be more than reluctant to upgrade just to > obtain a single but important bug fix. > > It is an annoyance that you still need a binary distribution before being > able to build ROOT from cvs (or even from cmz for that matter). The two > reasons why the binary distribution is really needed are the dependency on > a specific makedepend and the need, on some platform, for additional libraries > (libXpm.a). The source of makedepend will also be available from the > same cvs repository, so this is not necessarily a show-stopper. The Xpm > library could also be added to the cvs repository but either a mechanism > to conditionally download them need to be in place, either each distribution > will be at least 400K larger, even for platform that do not need the extra > libraries. > > In the longer term, I agree that ROOT need to move to an even more open type > of repository. This is because the current model will only work > as long as Rene and Fons continue to work full time on the ROOT project. > As soon as [if] their availability decreases, the fact that only the two of > them have write access to the main repository might become a problem for > maintenance purpose. I think movement toward this more open scheme is > more important that the exact solution implemented (cvs vs. BitKeeper) as long > as more a modern, easy, open way for many developers to contribute. > > Philippe Canal, Fermi National Laboratory, > (pcanal@fnal.gov) > > > > -- Org: CERN, European Laboratory for Particle Physics. Mail: 1211 Geneve 23, Switzerland E-Mail: Fons.Rademakers@cern.ch Phone: +41 22 7679248 WWW: http://root.cern.ch/~rdm/ Fax: +41 22 7677910
This archive was generated by hypermail 2b29 : Tue Jan 04 2000 - 00:43:38 MET