Source packaging arrangements (was Re: Debian vs. BOGUS)

[bhogan@bedlam.rahul.net (Bill Hogan)]

    [ short answer ]

"IJ" == Ian Jackson <iwj10@cus.cam.ac.uk> writes:

  IJ> Bill Hogan writes ("Re: Debian vs. BOGUS "):
  bh>> As I see it, Debian users have the following situation:

  bh>> [ diagram deleted ]


  IJ> No, the situation is as in my diagram below.

    [...]

    [ diagram deleted ]


    Ok, thanks.

    I was assuming a situation like this:


        original_materials
                |
                V
              apply <--------------- debian_changes
                |                          |           DEBIAN
    ============|==========================|=====================
                |                          |           USER
                V                          V
           debianized_materials ------> unapply
                                           |
                                           V
                                   undebianized_materials



    I now take it the situation is more like this:


        original_materials ----------------+
                |                          |
                V                          |
              apply <--- debian_changes    |
                |                          |
                V                          V
       debianized_materials ------------> diff
                |                          |           DEBIAN
    ============|==========================|=====================
                |                          |           USER
                |                          V
                |                   debianizing_diffs
                |                          |
                |                          V
                +----------------------> unapply
                                           |
                                           V
                                   undebianized_materials



    Here is the problem I have with this scheme.

    The only way I can be sure my undebianized materials are identical
to the original materials from which my debianized materials derived
is by comparing my undebianized materials to the original materials
from which my debianized materials derived -- that is to say, by
comparing my undebianized materials to the original materials that
went into making up the debianized materials from which I extracted my
undebianized materials.

    However, the original materials that went into making up the
debianized materials from which I extracted my undebianized materials
are not available to me in the present scheme.

    Therefore, I cannot in the present scheme be sure my undebianized
materials are identical to the original materials from which my
debianized materials derived.

    Maybe Debian should consider giving its end-users the option of
cycling their undebianized materials back through the loop, starting
from the Debian end, where it says `original materials' in the diagram
preceding this one, to verify that the same undebianized materials
come back.

    Such a procedure should of course be blindfold in the sense that
except for setting a switch to remind itself this was not a production
run Debian should generate the requested `undebianized materials' file
exactly as if the request had originated in-house and only at the very
end consult the flag it set earlier and on that basis compare the file
it generated to the file sent in by the user and e-mail a summary of
the differences between the two files back to the user.


    Cordially,
     Bill