On Sun, 5 Apr 2020 16:06:33 +1000 Daniel Kasak <d.j.kasak...@gmail.com> said:
> On Sun, Apr 5, 2020 at 5:23 AM Marc MERLIN <marc_...@merlins.org> wrote: > > > > I'm upgrading to a new laptop (thinkpad P73). > > > > It has a 100Wh battery, but it's also battery hungry, so one of my goals > > is to limit battery usage when using the intel driver only (I'll turn > > off the nvidia chip). > > That said, I'll still be running the usual X apps, from gthumb, to > > google-chrome, to arduino, to xmms, gvim, and so forth. > > > > Is it worth my looking into using Wayland and the lasted E I can find > > packaged for debian (sorry, still not interested in the build your own > > from TOT rabbithole :) ), or am I going to be better off sticking with X > > for now? > > > > Thanks, > > Marc > > I build E from source, with X and Wayland support, and run on a bunch > of devices, including Intel CPU / Intel GPU and AMD CPU / AMD GPU. For > regular use, I'd say stick with X. Here are some gotchas for me with > running under Wayland: > > - no external display support - only your laptop's LCD will work - > https://phab.enlightenment.org/T4501 yup. patches queued but they have reliability issues across drivers. i think i narrowed it down to a design decisions ensuring this would happen and that design decision needs to change inside the code in evas. > - usability issues with gtk apps - > https://phab.enlightenment.org/T8295 - I develop some gtk apps that > are affected by this, so it's a non-starter for me just with this bug toss this on the subsufaces bandwagon. those need work. > - usability issues flipping apps to/from fullscreen mode - > https://phab.enlightenment.org/T7663 just smells of a code path to debug and fix. > - high cpu usage ( E uses 100% of 1 core ) when running firefox ( I > haven't raised a bug for this yet, but I assume it's a known issue ) that's bad... firefox in x or wl mode? > - slightly more unstable than E on X that will take time and effort to just weed out the bugs. > - restarts kill X apps ( when combined with the above, this is painful > ) - known issue, apparently no wayland protocol needed to support > graceful restarts this ... is a core problem. a restart of e is re-executing e. that means sockets are lost - xwayland as a client is lost thus all x apps are lost. actually gtk apps in wl mode would be too. efl apps can survive because they reconnect nd use an uuid assigned by the compositor to indicate they are the same client again and e restores their state. other toolkits need to adopt the same protocol to get the same treatment. technically xqayland could do this too. it'd require patching it. we have enough other things like above and below to work on anyway before worrying about this... > - mouse/touchpad usability - no config for sensitivity, mapping mouse > buttons, scrollbutton, etc yeah. need the config dialogs and the libinput swizzling. > Having said that, E itself runs much smoother on all my systems > running in Wayland mode. Memory usage is lower ( not that I care too > much, but if you have limited memory I guess you would ). I give it a > go every couple of weeks to see how usable it is ... but all of those > above points for me combined make it less appealing that running on X > at this point. YMMV. my tests of e wl mode are mostly on my rpi's and rockpro64 - not as a daily driver. it seems decent enough performance-wise there, so good to hear it's smooth for you too. :) > Regarding the 'rabbithole' comment ... it's really not that hard to > build from git sources. > > Dan > > > _______________________________________________ > enlightenment-users mailing list > enlightenment-users@lists.sourceforge.net > https://lists.sourceforge.net/lists/listinfo/enlightenment-users > -- ------------- Codito, ergo sum - "I code, therefore I am" -------------- Carsten Haitzler - ras...@rasterman.com _______________________________________________ enlightenment-users mailing list enlightenment-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/enlightenment-users
