On Mon, Nov 18, 2013 at 08:40:38AM +0100, Mark Trompell wrote: > So we're down to the crossroad I feel, where we have to quickly decide > where we are heading. Do we want do revive foresight and jump on the > fl3 train or do we want to go on with our lifes and let foresight die.
I've been silently pondering this thread for a while. I apologize for being so quiet, but I didn't want to be too hasty. I still think that my old proposal (as Rune referenced) is a general good start. I'd suggest taking bite-size pieces would make it easier to get to something that works: 1. Import Fedora 20 (it is just about to be released) using mirrorball. Just do a binary import to start with. We might be able to start small by starting with one of the smaller "spins" and then add more packages as we go, in order to make the pieces more bite-sized. 2. Optionally, developers who would like to explore rebuilding from SRPMs with mirrorball could start a track doing that work, bootstrapping from the binary import, in a different repository. 3. Building Foresight as a conary-native layer that can run on top of either the binary import or the bootstrapped source build, that has only things different from the underlying platform, would let Foresight return to its roots. 4. Import newer versions of Fedora as they are released, and as they are ready, move the Foresight layer on top of them. Here's my problem: I know I will personally have limited time available to participate in this process. So I would suggest doing this only in concert with developers now at SAS who were previously at rPath. In order to make this proposal, I'd need to know who of the Foresight developers would be willing to run the import process, noting and investigating failures in the import process, and working through updates and fixes with those developers. To propose this internally, I'd need to know who was volunteering to work on the import process. So: If you care enough about this that you'd be willing to help drive the import process and help isolate bugs in the import process so that we can build on top of an existing base, please respond, either to the list or privately to me, so that I can propose this internally at SAS. Just being very clear: this is expressing my own ideas, and does not represent any views of my employer. But if we (Foresight) have enough commitment to working on this, I'm willing to propose it internally (SAS) and then follow up with the volunteers regarding the result. _______________________________________________ Foresight-devel mailing list [email protected] https://lists.foresightlinux.org/mailman/listinfo/foresight-devel
