Jaldhar H. Vyas wrote:
On Tue, 1 Mar 2005, Hugo Vanwoerkom wrote:I see your point. But out of curiosity, how many organizations are there like Debian? IM 1000+ DD's spread out over the world without a physical place, putting together a fairly sized software project, but always facing an oponent like Microsoft. I worked for IBM for 28 years and we sat together in buildings, but that isn't the case here. Maybe it is the "virtuality" what is "wrong".1000 developers sounds impressive but most of them just maintain a few packages. The vast majority of "infrastructure" work is only done by a relatively small number of people. The shear number itself causes inefficiency. (See "The Mythical Man-Month" for the classic study of why just throwing more bodies at a project doesn't make things go faster.) Companies can ask employees to work 60-80 hours a week as release time approaches. Debian can't ask that of volunteers who have real life comittments, might be demotivated, working on something else etc. The great distance and loose contact between DDs causes entropy in communication which further slows things down. These are the real reasons Debian is so slow to release. The "too many architecures" reason is a red herring as is "we are perfectionists" (At this point most DDs are pretty sick of the delay too.) Luckily all these problems are known and being addressed. * A revamped new maintainer process which helps make sure DDs have a baseline of skills and greater understanding of the goals and methods of the organization. * More joint-maintainership for large (or merely time-consuming projects) * More face to face meetings. * Greater willingless to take over packages from missing in action maintainers And I hope the new DPL will take action to remove the last few bottlenecks in positions of authority. Debian is a unique experiment. We've gotten some things wrong but over all we've gotten more things right so you can be confident we'll get over this hump.
I think a good point was brought up here. There is strength in numbers. We need to nurture new talent so that more Debian developers are born. The process is rigorous (as should be) but in the end how many new developers do we see each year? Perhaps forming classroom instruction seminars would help educate and promote new maintainers. This could be done with relatively low cost (look at the LUGs). Why couldn't we form similar groups where potential developers/maintainers could go and get face to face help from those in the trenches? I think projects like mentors.debian.net are one key to Debian's success. The problem seems to be that the developers/maintainers are so busy they rarely have time to provide in depth help to those trying to learn. In the end, if the time was taken it would ultimately increase the number of developers and reduce the workload of the mentor. -- "Imagination is more important than knowledge." - Albert Einstein
Attachment:
signature.asc
Description: OpenPGP digital signature