MJ Ray wrote: > On 2004-07-13 21:39:31 +0100 Francesco Poli <[EMAIL PROTECTED]> wrote: > >> I agree. Focussing on packages only would require too many analyses, >> indeed. > > Are you claiming that "this package fails to follow for the same reasons > as that one" requires more analyses than analysing the licence and then > "this package fails to follow because it is only under licence L"? I > remind you, we still should check the packages if asked. In fact, I > think if there are n packages under some combination of that licence, we > do at most n analyses if we mainly analyse packages and n+1 if we > analyse the licence first. Marginal, but possibly still significant, for > some.
Don't forget that once the license summary is done, the package summaries will generally be trivial, so saying "n" and "n+1" is inaccurate. If you are going to ignore constant factors, then you might as well say that both approaches will require O(n) summaries. :) If you use the summary of the first package with a given license as the summary for that license, then you may end up working around additional baggage in the summary that was specific to that package. If you start from a license summary, you just need to ask "how does this differ from the general case", not "how does this differ from <insert possibly obscure package here>'s case". >> [...] would be a waste of time if we had to review the >> same licenses again and again or to dig in the archives to recall if >> some old package in a similar situation was judged free or not... > > This is where we are at the moment. I thought the summaries were an > attempt to reduce the digging, but they seem to have drifted. How so? - Josh Triplett
signature.asc
Description: OpenPGP digital signature
