Re: Proposal: user-visible list of divergences from upstream
On Tuesday 09 September 2008 14:53:16 Eugene V. Lyubimkin wrote:
> Ben Finney wrote:
> > Only by getting the package and unpacking it (as I'm sure you know,
> > the package can be unpacked and inspected without installing it).
> Both these methods require:
> 1) knowledge about where to look for Debian changes (Debian users should
> know about README.Debian, but non-Debian?)
> 2) downloading some stuff
> Ok for geeks and developers, but too expensive for others, imho.
I failed to see how adding an extra file you are suggesting
(debian/divergences or whatever) could be better than already exising ones
like README.[Debian|source] - it is basically in the same boat with regard to
above mentioned requirements. OTOH, it is more reasonable to maintain well
documented patches using their headings (note, the single place to change),
which could then be extracted and revealed by patch-tracking.debian.net or
whereever. I doubt there is anything more direct and easier for regular
users, since they would read exactly what has been documented in the relevant
patch created by the developer, and would need only a www browser or text
editor, where the developer would has a single place to worry about.
> > Are you proposing that, in addition to the changelog and the
> > README.Debian and the NEWS.Debian and the package control files, that
> > there should be *yet another* place where the package maintainer is
> > expected to duplicate information on what they've done to the package?
> README.Debian contains notes about important changes that made in
> Debian's variant of package for a long time of package' lifecycle.
> NEWS.Debian is especially good for upgrading. Changelog is for
> developers and geek users as it contains developer stuff. Patches is
> almost purely developer stuff. In my view, all these files do not cover
> "actual user-oriented important divergences from upstream". Only in my
> view. I understand that this proposed info will interfere with above
> mentioned one in Debian documentation files and will require some
> additional time to maintain.
I believe that such data duplication would surely lead to discrepancies at
some point, which would add even more confision to the reader.
pub 4096R/0E4BD0AB 2003-03-18 <people.fccf.net/danchev/key pgp.mit.edu>