I believe that a clear division is being formed between
projects that are moving forward with clear regular releases and a tight focus (best example I can think of is Ubuntu) and those that are rambling on and may one day release but in the meantime lots of new stuff is being added far too many good examples Debian comes instantly to mind (although maybe some progress is being made see LWN subscriber only until 24th original email on gmane).
I suppose it should be obvious but this is such an easy mistake to make (and I know I have done it). Get into a machine like release crank out mode and projects just go better, it is a positive spiral, start letting releases slip and you end up wasting so much time maintaining multiple versions that progress seems to disappear.
The big vision is needed but far better is the ability to crank out incremental innovation on a regular and predictable release cycle, you can get to the same point so much faster. The bigger the project the more important this is. I think we can see Java getting this itself and within the java community many get it but are struggling to get it working in practice (thinking of OJB for example).
The problem is that it is much easier to get the regular releases working when a) the tight focus is there and b) the release is easy. This is a big score for internet based applications where the release is fully controlled and where there is a clear interest by the hosting organisation to keep focus so that the project is manageable see the links inc "42: The future of google?" for how this is working for google and against Microsoft.
All promoted by How to delay a Zope release.
Comments