nachoig wrote:PulseAudio is here because ALSA is not a sound server.
Actually PulseAudio is here to unify the Linux sound libraries and provide an unified API, this is why it has an extensible plugin architecture with numerous backends. Which doesn't make sense in the first place since it aims to unify audio libraries fragmentation by providing one more fragment.
Additionally, a sound server itself doesn't make much sense, as it only increases layers and sound latency and sound programming complexity, as you have to keep in mind anything you do is asynchronous, moreover any feature provided by the PulseAudio server could be achieved extending ALSA itself, or even OSS, and this is demonstrated by OSS4.
systemd is not only about sppeed. systemd is more than this
That is an interesting point made by systemd creator himself just recently, I guess that's why he spent about 3 conferences explaining how fast systemd is compared to other alternatives, without focusing on many other relevant arguments.
it's capable to track all daemons, its unit files are much better than SysV's scripts, these unit files aren't distro-specific, it has excelent administrative tools.
Its features have been implemented on many other less known (and advertized) init systems, such as s6, runit, upstart. Among them s6 is very interesting, since it is completely SysV compatible. Whether or not shell scripts are better is not that easy to say, as shell scripts are much more flexible, you'll also notice that many systemd unit files resort to invoking shell scripts to start daemons, which kinda defeats the purpose of having them.
No, I can't compare this with OpenRC. OpenRC is only an inprovemente running on top of SysVinit, but it falls in the limitations of SysV... Also, parallelisation in OpenRC doesn't work properly.
Parallelization in OpenRC does work, as cgroups
support does, the fact that rc_parallel isn't recommended because synchronous boot is safer doesn't mean that it isn't working.
X: X is bloated, and it has a lot of non-functional code and needs a lot of ugly workarounds.
The core X protocol hasn't been updated in 30 years with the excuse "We are not the X consortium", thus it has only been extended and worked around, I see little point in complaining about X when nobody done a single thing to improve it. I'd actually say it's a pretty good and solid standard if it has been able to survive and be extended all these years.
Regardless X, this doesn't explain why Mir has been announced despite Wayland doing exactly the same thing, little plausible explainations are available for this...
Anyway, as long as nothing gets forced on users, anyone is free to use and support whatever is better to him/her.