Initial observations or Brandon's proposal: 1: Please separate the "whereas" from the actual proposal. People might agree with the proposal, but for different reasons, and might not want to seem to be signing on to the whereas clauses--and reject it for that reason alone. I assume you don't care if everyone has the same reasons for agreeing with you, in which case, please separate the proposal itself from the reasons you favor it.
2: The 16K limit should be checked against existing packages before we say that's the number. It being universally agreed that we aren't upset by the cases we know about, we should check and make sure they don't run afoul of any new guideline. 3: I would *MUCH* prefer a limit that involved some kind of proportionality to one that has a fixed size. 4: You are attempting to be "legal" and cover all cases. That's already outside the spirit of the DFSG. Please resist the (understandable) temptation to try and define things in such a way that nobody ever need exercise any judgment or understanding in the future. Elaborate definitions of like "instruction code for a computer" are things that are not helpful: we all already know what that means, and it's a purely internal guideline, so nothing is served by making it too rigid. We should never create internal guidelines which have the effect of perhaps blocking us from doing the right thing at some future date. 5: Your notion of a "work" is unclear. Are we talking about manuals? Pretty much no manual is "instruction code". Or packages? This is important if and only if you want to maintain the distinction between the two sorts of cases, and there is no motivation I can see for doing that. 6: I am *still* unclear why we need a policy. I would prefer to avoid making policies until and unless they turn out to be necessary. In the present case, it would only turn out to be necessary if we envisioned frequent arguments among developers about whether given packages should or should not be added. (Having only one argument, should it happen, is no matter; we had that about KDE and it was not necessary there to make any formal interpretive policy statement--in general, we should avoid making formal policy statements, period.) 7: Here is the sort of policy I would happily support, presuming you can explain why, exactly, we need one at all: "Some documentation or other matter in Debian packages is sometimes distributed under licenses that do not permit modification or the distribution of modified versions. When these portions are small relative to the size of the package, no harm accrues from distributing them as part of Debian. However, when it comes to actual programs of whatever sort, we continue to insist that modification always be permitted. And, since changes to a program necessitate changes to its associated documentation, any portion of documentation which might need to be changed to correctly describe a change in the program must also permit modification and distribution of modified versions." That seems to meet all the requirements. Rather than ordering developers about what consitutes "small", I think we should trust that developers are reasonable, and only if a problem seems likely to loom should we go about trying to be more specific. Thomas
