[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: systemd



On Thu, Jul 26, 2012 at 06:30:23AM +0200, Ralf Mardorf wrote:
> To get off the track: Not only Poettering's stuff is "experimental",
> also the switch from nv to nouveau IMO is insane.

Not at all.  IIRC, nouveau supports KMS while nv does not.  nouveau
therefore integrates with modern X11, drmfb and other technologies.
nv does not.  Now, nv might provide a better driver experience right
now, but don't let short term inconvenience block major long-term
improvements.  Look at the state of where we are now, and where we
should go in the long term:

Right now, if you use either the nVidia nv or AMD fglrx drivers,
you get a complete current OpenGL stack and great hardware
acceleration.  The downside is that each has a completely
independent OpenGL stack, which both have bugs, and which don't
play well with the rest of the system.  They both break often
because they don't keep up to date with current development of
kernel and X11 interfaces.  They are both known to cause kernel
instability due to their insane designs.

The open drivers are only just starting to offer OpenGL 3.0
support, and are not as complete and not as performant.  However...
we have a unified OpenGL stack, and all the drivers are kept
up to date with current kernel and X11 interfaces.  Unlike the
proprietary drivers, "old" graphics cards are not removed within
a short timespan.  They can make use of the kernel infrastructure
like KMS, DRI/DRM, GEM/TTM etc. while the proprietary drivers don't
play nicely.

In the future, we want to be in the situation that all drivers
use the same OpenGL stack and are all free.  It means as a developer,
you'll be able to rely on the presence and functionality of a
certain OpenGL baseline (e.g. 3.x, 4.x in time), while at present
the baseline is 2.x (you can't use higher level features without
only being usable with proprietary drivers).  The proprietary drivers
have no long term future--it's not like they have ever been truly
supported, due to the kernel tainting etc.  As a developer, I want to
be in the situation where I can just distribute an OpenGL-using
application and have it work out of the box; the proprietary drivers
will never provide that.  It's also a lot easier to develop for--see
recent comments from Valve et. al. regarding it being far easier to
optimise and debug given the source availablity.

The last two cards I bought were AMD/ATI HD  4850 and 6970 cards.
They have only ever been used with the free drivers.  Stability
is great, 3D support is good, and I have no complaints.  I
purchased these specifically because of the fact that it has
good libre driver support.  In time, we'll get better performace
and more features, but for now we have a Free and functional stack,
albeit not as slick as the proprietary offerings.

> I'm sceptical, but to be fair, I never tested systemd myself.

Criticism is better placed when it is informed criticism.  I would
highly recommend giving it a try.  While I disagree with a number
of systemd decisions, many of the core concepts are solid, and this
type of init system is definitely the future.  Even if we don't
default to systemd, reusing its concepts will be highly beneficial.


Regards,
Roger

-- 
  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux    http://people.debian.org/~rleigh/
 `. `'   schroot and sbuild  http://alioth.debian.org/projects/buildd-tools
   `-    GPG Public Key      F33D 281D 470A B443 6756 147C 07B3 C8BC 4083 E800


Reply to: