After listening the packagekit talk at FOSDEM, Fabio and me started to think about using tracker for supporting distribution packages system. I'm going to talk with sabayon's entropy package manager in mind, but it may also apply elsewhere like on debian, or at a higher level, to packagekit.
We want try to use tracker to store information about available apps, with available apps i mean installed software [that miner-fs already index] but also installable apps [read packages present in the distribution repository] For doing this we are going to write a small miner, that monitor repository updates and the insert into tracker new software, update old info and so on. Looking at ontology we can already use nfo for some thing, for specify if an apps is installed we can use tracker:available [or you have some better idea?], and we can link other info such screenshot, video, comment, etc to the application [what relation we should use?]. Maybe we can also need some specific ontology, for all metadata we have in the db, but for a first step it isn't necessary. Now think on the great part of this things. If we define a standard ontology to use for this use case, packagekit can become a simple tool that query tracker, and show user info on what software are installed, display search result for now awesome program and so on. And any distro should only write a simple miner, and it became packagekit friendly. Or maybe we can use fster to provide a navigable hierarchy of apps categories, so a user can click on his desktop icon fresh_software, enter in the game folder, look for his preferred game, click on the package-file and just install it!!! i'm really excited about this :) [and this can became cross-distribution without any effort!] What you think guys about this? suggestion? question? Michele -- michele tameni _______________________________________________ tracker-list mailing list [email protected] http://mail.gnome.org/mailman/listinfo/tracker-list
