Re: Reverting to GNOME for jessie's default desktop
Jordi Mallach writes ("Reverting to GNOME for jessie's default desktop"): > It's been around 9 months since tasksel changed (for real) the default > desktop for new installs. At the time of the change, it was mentioned > the issue would be revisited before the freeze, around debconf time. Fascinating as this discussion is, I think it is at risk of becoming too much of a time sink. I think that it would be useful to have some authoritative guidance from those in Debian who are responsible for this decision. AFAICT that is the tasksel maintainers. So I would appreciate it if the tasksel maintainers would let us know: Do you intend to review (or are you reviewing) the decision taken in July 2012 [1] ? If so, is this discussion here on -devel useful ? If it is useful, what questions should we be focusing on ? Ian. [1] http://anonscm.debian.org/gitweb/?p=tasksel/tasksel.git;a=commit;h=2a962cc65cdba010177f27e8824ba10d9a799a08 -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/21482.8088.55315.575...@chiark.greenend.org.uk
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
I enjoy the way you keep ignoring the relevant points, memory usage and performance regressions. And the way you benchmarked gnome against gnome. How about warsaw on xfce on the same hardware or your benchmarks pretty much show nothing except that your 'slight performance increase when using gnome-shell' is pure fabrication. On 12 August 2014 13:20, Josselin Mouette wrote: > Le mardi 12 août 2014 à 13:12 +0100, Anthony F McInerney a écrit : > > Virtualbox Results (no guest drivers installed) > > Glxgears is not a relevant 3D benchmark. > > But the funniest thing is that you did this test without any 3D > acceleration, which is not representative at all of most real-world > computers. > > Thanks for making the point that with llvmpipe, GNOME is perfectly > usable on a machine without 3D. > -- > .''`.Josselin Mouette > : :' : > `. `' > `- > >
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
> > We happen at work to have users with very important needs of 3D > resources, so one of my colleagues conducted some performance tests with > and without a compositor (the compositor being GNOME 3). > > It turns out that with a recent adapter, 3D applications are running a > small bit faster under GNOME, and that’s probably because it saves your > graphics card the pain to switch from 2D to 3D contexts. > > Virtualbox Results (no guest drivers installed) Gnome: glxgears 1267 frames in 5.0 seconds = 253.150 FPS 1197 frames in 5.0 seconds = 239.372 FPS 1174 frames in 5.0 seconds = 234.753 FPS 1142 frames in 5.0 seconds = 228.005 FPS 1201 frames in 5.0 seconds = 239.898 FPS 1217 frames in 5.0 seconds = 243.075 FPS 1194 frames in 5.0 seconds = 238.475 FPS Gnome-Classic: glxgears 1150 frames in 5.0 seconds = 229.772 FPS 1267 frames in 5.0 seconds = 253.212 FPS 1240 frames in 5.0 seconds = 247.875 FPS 1272 frames in 5.0 seconds = 254.374 FPS 1314 frames in 5.0 seconds = 262.673 FPS 1279 frames in 5.0 seconds = 255.684 FPS XFCE: glxgears 2110 frames in 5.0 seconds = 421.982 FPS 2290 frames in 5.0 seconds = 457.800 FPS 2241 frames in 5.0 seconds = 448.046 FPS 2279 frames in 5.0 seconds = 455.636 FPS 2191 frames in 5.0 seconds = 438.094 FPS 2246 frames in 5.0 seconds = 449.114 FPS 2218 frames in 5.0 seconds = 443.590 FPS Please note xfce was installed on the same vm, that gnome / gnome-classic was on. It was therefore using much more memory, (it seems GDM + other gnome services are causing memory use, i will be testing that shortly). And here i will make a comparison. ahahahaha. That was it. Moving on. ;)
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
Le mardi 12 août 2014 à 13:12 +0100, Anthony F McInerney a écrit : > Virtualbox Results (no guest drivers installed) Glxgears is not a relevant 3D benchmark. But the funniest thing is that you did this test without any 3D acceleration, which is not representative at all of most real-world computers. Thanks for making the point that with llvmpipe, GNOME is perfectly usable on a machine without 3D. -- .''`.Josselin Mouette : :' : `. `' `- -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/1407846059.26277.455.camel@dsp0698014
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
On 12 August 2014 09:51, Wookey wrote: > > Could you do MATE too please? > > MATE: (with mate-desktop-environment-extras) free ^[[C total used free shared buffers cached Mem: 506756 397480 109276 7096 58820 166076 -/+ buffers/cache: 172584 334172 Swap: 392188 0 392188 The ctrl characters came with it apparently. And as an extra, Gnome Classical: free total used free shared buffers cached Mem: 506756 498240 8516 2452 5972 71360 -/+ buffers/cache: 420908 85848 Swap: 392188 16292 375896
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
Le mardi 12 août 2014 à 03:03 +0100, Anthony F McInerney a écrit : > I had stated previously XFCE had started showing memory usage similar > to gnome. This has quite obviously changed. I was wrong, and i'm > posting it as a correction to my statement. You’re comparing apples and oranges. These memory usage comparisons are only useful at feature parity - which doesn’t exist, since different environments have different paradigms and different feature sets. > How much RAM does your browser use? > Lots, which is why i prefer my DE not to eat it all. When the browser uses 1 GB while GNOME (including evolution and many other running applications) uses half of that, I don’t think you need to look for memory savings in the desktop. You need to buy more RAM and that’s all, because browsers won’t suddenly stop needing their gigabyte. > If you can't run GNOME because you don't have the system specs > to run a modern desktop then you can select XFCE/LXDE in the > installation menu. But let's be fair, those people are a > minority. And a default should fit the needs of the majority. > Ahh good you have statistics for that. Please link them, or quote and > cite sources. I just had a look at an online hardware store. Out of their 682 laptops and 332 desktops: * 1 model has 1 GiB * 48 models have 2 GiB * 470 models have 4 GiB * 495 have 6 GiB or more Which means 0,1% of the machines you can buy are not able to run a web browser anyway, < 5% are more than enough for a full-fledged GNOME+web browser, and all the rest are very comfortable with anything you can run under Linux. > Some people like the 'basic 3d acceleration' for other things, so not > only do you want me to sacrifice my RAM to all powerful DE, but also > my GPU? How kind of you ;) We happen at work to have users with very important needs of 3D resources, so one of my colleagues conducted some performance tests with and without a compositor (the compositor being GNOME 3). It turns out that with a recent adapter, 3D applications are running a small bit faster under GNOME, and that’s probably because it saves your graphics card the pain to switch from 2D to 3D contexts. -- .''`.Josselin Mouette : :' : `. `' `- -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/1407838921.26277.435.camel@dsp0698014
Re: [Pkg-xfce-devel] Reverting to GNOME for jessie's default desktop
+++ Anthony F McInerney [2014-08-12 00:02 +0100]: >XFCE: > > total used free sharedbuffers cached >Mem:506756 362468 144288 6568 22756 179264 >-/+ buffers/cache: 160448 346308 >Swap: 392188 0 392188 > >GNOME: > > total used free sharedbuffers cached >Mem:506756 500360 6396 1948840 37724 >-/+ buffers/cache: 461796 44960 >Swap: 392188 66672 325516 > >LXDE: > > total used free sharedbuffers cached >Mem:506756 316504 190252 8500 18920 149812 >-/+ buffers/cache: 147772 358984 >Swap: 392188 0 392188 > >KDE: > > total used free sharedbuffers cached >Mem:506756 499724 7032 6772 10516 109760 >-/+ buffers/cache: 379448 127308 >Swap: 392188 21632 370556 Thanks for this, interesting. Could you do MATE too please? Wookey -- Principal hats: Linaro, Emdebian, Wookware, Balloonboard, ARM http://wookware.org/ -- To UNSUBSCRIBE, email to debian-project-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: https://lists.debian.org/20140812085136.gm7...@stoneboat.aleph1.co.uk