[EMAIL PROTECTED] (Thomas Bushnell, BSG) wrote on 04.09.02 in <[EMAIL PROTECTED]>:
> Russ Allbery <[EMAIL PROTECTED]> writes: > > > The arguments that have been presented that say that requiring file > > renaming is an infringment on the freedoms guaranteed by the DFSG are > > certainly reasonable ones and I can find much in them to agree with, but > > the DFSG really *aren't* clear on this point, and a ruling on the subject > > does not just obviously fall out of what's already there. > > DFSG says that you have to permit modification. (By patches or > directly.) That is violated by a rule like "if you modify this, you > must chant the kama sutra" or "if you modify this, you cannot name the > output file foo.bar". It is not actually clear to me that this is violated by any of the example rules. > The reason the latter is crucial is because it > is an *operational* matter for the software, changing such things as > APIs is exactly why we want the right to modify files. And we usually get upset if a library changes its API without changing its soname or versioning its symbols. *Because* this is an operational matter. > > (As an aside, once Debian reaches some sort of general conclusion on this, > > it would be really nice to add that to either the DFSG or some supporting > > material, since this has come up repeatedly for years and this exact > > argument happens every time.) > > Eww, no. The current method is actually better. It takes time and > patience, but the result is an ever-growing cadre of people who have > thought it out and talked about it, and so much more deeply understand > it than if they just read it in some document. This is insane. MfG Kai
