From: Johan Nilsson (johan.nilsson_at_[hidden])
Date: 2004-10-30 13:40:47
"Andreas Huber" <ahd6974-spamgroupstrap_at_[hidden]> wrote in message news:cluamk$sa6$1_at_sea.gmane.org...
> Johan Nilsson wrote:
>> Could the NIH syndrome be diminished by easier accepting more people
>> into the actual development process for a particular library? Ok, so
> For me the NIH syndrome always implies a good amount of non-technical
> arguments, which are almost absent here. So, I don't think that the
> boost community actually suffers from NIH. It might suffer from
> over-perfectionism. However, I believe this is exactly the reason for
> the exceptional quality of the libraries and would therefore not want to
> change anything.
>> I'm mostly a lurker here, but even so I believe I've seen people from
>> time to time declaring their willingness to participate in developing
>> a new library, without getting some real response. Take them in, let
>> them be a part of the effort, and maybe the NIH problem will be
> I'm rather skeptical about that approach. Groups of similarly capable
> people are usually not very good at *designing* a library (unless
> there's a boss who has the last word on all the decisions).
So you mean that a single capable person beats a group of capable people? I realize that having a group of people trying to work out a design across the internet might not be feasible, but generally I still think that having a group of people together in the same room, preferrably at a white board, beats a single person. Having someone responsible that has the last word might be necessary anyway, but preferably only as a last resort. That should be possible here as well.
It is much too easy locking yourself into a single mindset when you work alone at the drawing board. This is why I heartily supports the idea of discussing the design openly, collecting the feedback and ideas - before putting so much time into developing and refining the library that it feels like so much work has gone into the library that one doesn't wan't to give up the design. Just so that you don't misunderstand me, I know I discussed with you quite a lot about the fsm library but I'm definitely not pointing fingers anywhere specifically here. People must be allowed to defend their design decisions as well as people must be allowed to challenge them, IMHO.
> However, I
> think groups are unbeatable at uncovering design-flaws.
True, OTOH it is much easier to find a design flaw than to generate a good, solid design.
> So the best
> designs are initially often invented by an individual and the input
> later given by a group (users, boost-members, etc.) should then be used
> to refine until both sides are satisfied. Unfortunately, this puts most
> of the burden on the library submitter but I think it is the best
> process if you aim for high quality.
Perhaps. I recognize that my point of view is being out-numbered here.
The process of taking the idea as early as possible to the newsgroup to collect requirements and implementation ideas still feels like a good idea. After this the "submitter" would stand a better chance of generating a good, perhaps easier accepted design and implementation.
Best regards // Johan
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk