Attached are patches that make yet more use of threads to allow marking the storage spoke selector as insensitive until storageInitialize is done. Once it's done, the spoke selector is marked as sensitive via a callback passed through in a kind of messy fashion. I know at least Will is going to object to this approach, based on talks in #anaconda. He'd prefer that I set it up so you can immediately go into the storage spoke and devices are shown as they are detected. I think that's a good idea, but I have chosen to not do so for the following reasons: (1) It would require changes to storageInitialize and DeviceTree that I don't believe we have time or manpower for right now. (1a) We can always go back and add this kind of fancier, on-demand behavior later. I don't think i've done anything to prevent that. (2) Storage probing is done in multiple passes and builds up as a tree. I think we'd run into a problem where component devices of more complicated structures are shown as intermediates. I'm thinking the drives that are components of a multipath or RAID set here. (3) I had more reasons, but I cannot recall them. Anyway, I expect very few people to ever see this behavior no matter how fancy it is. Storage probing starts very early on, even before the UI is up, which provides a lot of time for probing the handful of local disks that makes up the vast majority of users. Much fancier setups will definitely take longer, and I believe they'll be fine with waiting a bit. I must admit I'm a little worried about this expanded use of threads, but I can't see anything beyond storage (and perhaps install source, which needs to know storage info) needing to do this. I would at least like to discourage use of the stuff in these patches. If everyone thinks I'm not completely crazy here, I'll work on making the notification shinier. - Chris _______________________________________________ Anaconda-devel-list mailing list Anaconda-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/anaconda-devel-list
