Re: [trinity-users] A Memory Comparison of Light Linux Desktops

On Thu, 18 Apr 2013 19:15:06 -0500 "Timothy Pearson" <kb9vqf(a)pearsoncomputing.net> wrote: I will try to show things like they look like from my perspective and may be you will find something useful. Some background: Although I have modern hardware in my possession, my favourite all-around machine is still an ASUS Eee PC 901, with an Intel Atom single core CPU running at 1.6Ghz, 2GB RAM, two 4GB + 8GB slow SSD drives. To give you a general idea: roughly it can be described as a top-tier PIII system on steroids (amount of RAM, seek time of the block devices). By the time KDE3.5 reached its EOL and was about to be dropped from Debian I already knew that KDE4 was not an option for me (way to heavy and buggy at the time), so it was obvious that sooner or later I have to abandon the ship, and in order to do so one needs to find out DE-agnostic alternatives to every single KDE suite application. And to achieve across-DE mobility I did. Some things, however, either had no alternatives outside KDE ecosystem, or these alternatives did not qualify in my own workflow. Here is some of them: kwin kpowersave krandrtray (it's 2013 and I have enough fingers on my palm to count all xrandr interfaces that actually do more good than the simple 'xrandr --auto' command; if I ever start to program for desktop, it's a good project to begin with) KDE notification framework (as a mean to hook a script to almost any conceivable event) kicker (believe me or not, it's hard to find a panel configurable enough to act, look and feel a bit outside of "fits everyone" point of view, so far only tint2 has qualified) mount manager This was probably the major points, but beyond that KDE/TDE have a long list of applications one could live without or find a substitute, but simply reluctant to do so. Back to my case, I just want a bloat-free, robust classic desktop from Debian stable main repository. There is nothing wrong with TDE except the fact that it is not there. Regular attempts to find a substitute just proved how good KDE3 series was and it was my own folly of taking all those features for granted. Now I almost feel like I'm giving "The grass was greener back then" speech, heh :). I'd like to share some of my thoughts on the "Qt4-bloat" topic. It is true, that by one reason or another transition to Qt4 resulted in noticeable increase in RAM usage. What I'd like to debate is the "speed" or CPU usage of Qt4 toolkit. Recently while evaluating TDE alternatives due to upcoming Debian Wheezy release I accidentally made a discovery which makes Qt4 look not as bad as it used to be in this particular aspect. I installed Razor-Qt, but accidentally forgot kwin4, and simply launched openbox instead. Overall snappiness just blew me away, still not as good as Qt3, but very impressive. For a moment I thought guys at Razor-Qt hit a jackpot: I installed KDE4 and observed how the very same set of KDE4 apps redraw itself noticeably slower. But then I started back Razor-Qt, it launched newly installed kwin4 and the miracle was gone. I tried kwin4 with minimal compositing and without, the results were the same: as soon as you launch 'openbox --replace' things were drawn a lot faster. That day I almost stayed with KDE4, but I guess I have a bit too many custom kwin rules to walk away like that. For someone not so kwin-bound this may be a solution. It's worth mentioning that I had set QT_GRAPHICSSYSTEM='raster' at the time and kwin4 memory usage hits ridiculous 65-70MB. TL;DR: In my opinion TDE is good because of minimal set of core applications which make it's a DE is both feature rich and resource-wise by today's standards. The only problem with it, that you can't find it in the main repository of today's major distributions. Much of the talk dedicated to slow redraw of Qt4-based desktops may be attributed to a kwin4 performance regression. Regards, Sergey.