On Thu, Dec 10, 2009 at 10:37:38PM +0200, Eugene V. Lyubimkin wrote: > > Well, even in that case, you _eventually_ need to split up the universe > > into actions/transactions/... or whatever the low-level package manager > > calls them. For instance, in the full state you will always have some > > (usually a lot) of the currently installed packages and you will > > eventually need to recognize that you will not need to do anyhow with > > them, because they are good as they are. > I don't understand how is external solver is related to inner-tool actions. Sure it is not. My point is that the package manager will have _anyhow_ at some point to compute a sort of diff between the current state and the one returned by the solver; hence it in fact make sense for the solver to compute that diff for the solver. > > If not, I can support a kind of protocol that I anticipated already: in > > the "preamble: " of the request to the external solver we add a > > "requested-format" or something such to choose among "new state" and > > "diff" styles. > Yes, I would like to have a way to request the target system state. OK, I'll then refine the protocol offering the two output formats. BTW, it would be very helpful if the apt-get (I mean, plain old APT) guys can voice their preferences on this: would you prefer a "new state" style of output or a "diff" style of output? Cheers. -- Stefano Zacchiroli -o- PhD in Computer Science \ PostDoc @ Univ. Paris 7 zack@{upsilon.cc,pps.jussieu.fr,debian.org} -<>- http://upsilon.cc/zack/ Dietro un grande uomo c'è ..| . |. Et ne m'en veux pas si je te tutoie sempre uno zaino ...........| ..: |.... Je dis tu à tous ceux que j'aime
Attachment:
signature.asc
Description: Digital signature