Version: unspecified (using KDE 4.5.2) OS: Linux Lately most distributions have done a fantastic job in improving boot time. Today, if we consider the total time required from switching-on the computer up to the moment when the desktop is ready to be used, we can see that the bottleneck has become the KDE startup time which takes typically more than 50% of the total boot time. Improving this will have tremendous benefits for all desktop users, and even more for netbook users! It will also put even more into light all the progresses that have already been done in improving boot time. Reproducible: Always
Bug 243277 seems to suggest that startup can be slow with loading Plasma applets. If you can isolate a different issue, please add a comment. You can compare start times from a fresh user (the initial loading will be slow because of ksyscoca building).
The bug you mentioned is another matter. I measured on a fresh user: the login takes 25s up to the sound notification. There are some projects whose goal is to boot linux in 5s. What I am proposing here is to have an equivalent goal: a graphical login in 5s. Is it feasible? What is needed to achieve it? What are the advantages and drawbacks? Wit the soar of small devices, it might be the good moment to address these questions.
*** This bug has been confirmed by popular vote. ***
Reducing startup time would go a long way to reduce the number of "KDE is too bloated" gripes seen in various end user circles. One reoccurring argument I see pointing in the direction of "it's good enough" is that users seldom need to reboot or logoff. This may be the case for a lot of individuals but there are are many situations which cause booting to be more frequent. For example, I have a few systems which suspend/hibernate just don't work right. I still want to conserve power so I shut them down when not in use. Another case would be users who share workstations and have their home directory on a network mount. In short, I don't believe this would be a case of optimizing where it's not needed. Moving forward, we need to figure out where we're spending our seconds. What instrumentation do we have that we can use to point blame? Can we make a "bootchart" of sorts for KDE loggins? Can we do more in parallel? Can we do less? (i.e. eliminate redundancies or avoid doing something that only satisfies a rare corner case use) Are there places we're using a sledge hammer where we could be using a screwdriver? I'm no KDE developer, but I'm going to poke around a bit and see what I can find in my free time over the next few days and post back when I have something that's hopefully interesting enough to make this bug report useful.
*** Bug 275133 has been marked as a duplicate of this bug. ***
New bootchart (without prelink/preload). dont know if it helps someone: http://img7.imagebanana.com/img/jawsh46a/kubuntu_natty_eeePCinspiron.png right eeepc / left inspiron
the situation is not changed with kde 4.9 and 4.10, gnome/cinnamon start in about half time than kde4, kde3 was very quick to start, this is a regression unfixed since 4 years, disabling kde services (http://ubuntuforums.org/showthread.php?t=1889034) doesn't improve the boot speed so much, please consider startup time improvement at least for kde5 ...
Configurability and modularity comes at a price. With the splitting of modules in frameworks, the startup time won't get smaller.
Uhm, this means no solution in kde5 too?
We had several failed attempts to improve startup time. With the continuous requests for more features and more configurability, any optimization on existing code is canceled sooner or later with new code. You can improve startup time by removing some functionality or merging multiple modules into a single file (e.g. a complete basic desktop into a single executable, like other environments use). Not sure if this is the right direction, given that you can start once, and simply use suspend/standby for subsequent starts. Nevertheless, the obvious "Patches welcome" conclusion applies here, too.
https://dantti.wordpress.com/2013/02/27/1-2-3-plasma/
*** Bug 324214 has been marked as a duplicate of this bug. ***
Startup time has generally improved a lot recently and I haven't heard any significant complaints about it in a while. See http://blog.davidedmundson.co.uk/blog/plasma-startup/