Hey, I'd like to discuss how we can store appstream data for applications in an xdg-app repository. Appstream data is what gnome-software needs in order to present applications before they are installed. The appstream data comes in several pieces:
- a single xml file that contains descriptions and other textual metadata for all the applications in the repository - icons for all the applications - screenshots Currently, Fedora ships the xml and icons in a single blob that gets installed together with gnome-software, while the screenshots are downloaded individually, from a different location. When I first looked at this, I thought I would treat the xml like the summary that ostree already has (just a plain file, not checked into the repository), since it is somewhat similar: metadata about the overall contents of the repository. But the prospect of downloading all the application icons with individual http requests doesn't look that appealing: essentially, we would do manually the same thing that ostree already does for us. Therefore, I've changed the approach to instead check both the xml and the icons into a branch with a fixed name. The code for that is here: https://github.com/matthiasclasen/xdg-app/commits/wip/appstream With this approach, gnome-software can simply pull this ref and get all it needs. And maybe more importantly, incremental updates to this data can take advantage of any delta download improvements ostree comes up with, as opposed to the Fedora approach, where we re-download everything every time a single icon changes. Screenshots are not included in this, and we would continue to treat them as separately downloaded items. Does this sound like a reasonable approach ? Matthias _______________________________________________ gnome-os-list mailing list [email protected] https://mail.gnome.org/mailman/listinfo/gnome-os-list
