On Wed, 2012-06-20 at 15:47 -0400, Chris Lumens wrote: > This came up in the meeting today, so here's a patch to start splitting > things apart so testing the UI doesn't result in your language changing. > Note that datetime and network spokes are not substantively included here. I think that the keyboard spoke is problematic as well. Having a text entry to test layouts, we need to change the X runtime configuration instantly after adding/removing/reordering layouts. On the other hand in cases of an image or live installations, we want to preserve the user configuration. So how to deal with this? Save the configuration in the begining of the installation and then restore it every time our window loses focus? This could be quite complicated per se and it would bring another problem -- desktop environments typically have their own layer on top of the X layouts configuration. Should we write a bunch of code dealing with gnome-session's layouts configuration, KDE's layouts configuration, etc.? I believe it could be better to just make the entry for testing layouts insensitive and do not change X runtime configuration when adding/removing/reordering layouts in the system-config-kickstart/live/image-install mode. But for this, the spoke needs to know in what mode it is actually running. The same goes for the datetime spoke. I believe everybody agrees, that setting time and date in the system-config-kickstart/image-install (and maybe live install as well) mode is a nonsense, so the part of this spoke should also be (in)sensitive depending on the mode it is running in. Starting/stopping chronyd.service when turning NTP on/off is the same case. (btw, the timezone switching is ok as it just sets the $TZ environment variable for the Anaconda process) So summed up, we need to get the information about the mode to the spokes (in a better way than testing flags.imageInstall and flags.liveInstall). That way, I can easily patch these two spokes to not ruin the environment when running on the live system. I'm also thinking about some "change_runtime_exec" function in iutil, that would actually not do the exec if running in the live/image-install mode. Having something like that could simplify the spokes' code. Comments are, of course, more than welcome. And once we agree on a way how to deal with these issues, I'll write patches. -- Vratislav Podzimek Anaconda Rider | Red Hat, Inc. | Brno - Czech Republic _______________________________________________ Anaconda-devel-list mailing list Anaconda-devel-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/anaconda-devel-list
