[Bio-linux-dev] Ubuntu Mir Concerns

Fri Jul 5 08:44:15 EDT 2013

Hi Jan,

I think you misunderstand what the changes actually entail.  The
"fallback" mode for legacy applications should be no less reliable than
the current XOrg X server.  The compositing X server on Ubuntu 12.04
already renders indirectly so you could reasonably regard it as a type
of "Mir light" where only the Compiz window manager is allowed to really
make use of the graphics (eg. to let you zoom in and out) but all other
applications have to talk through the clunky X protocol.  The move to
Mir/Wayland will allow applications to bypass the X protocol and talk to
Mir directly but applications that need X will continue to work as
normal.  All the code to make this happen will be taken straight from
Xorg and will work as before - we'll just lose the clunky driver
architecture and all the legacy extensions that nobody uses, or even
remembers, any more.

And regarding the abstraction used by authors of graphics packages, the
library interfaces do not change, it's just that the rendering code
lives in a different place.  The native rendering of 2D graphics
primitives in X is a hangover from the days when this was too much work
for the CPU on the server, and those days are long gone.  To remind
yourself what native X rendering looks like, start "xfig".  Any
application that doesn't look like that is already rendering with a
client library, and you've been using many such applications happily for
years.

Trust me - the end of the X server is both inevitable and desirable, and
nothing is going to suddenly stop working.

Cheers,

TIM

On Wed, 2013-07-03 at 15:14 +0100, Jan Kim wrote:
> Dear Tim,
> 
> thanks for your comments, here are a few further remarks (mostly
> technical and not really related to bioinformatics, I hope that's ok).
> 
> On Mon, Jul 01, 2013 at 03:15:05PM +0100, Tim Booth wrote:
> > Hi Jan,
> > 
> > Everyone in the know seems to be saying that the transition away from X
> > as the core Linux display server is "a good thing", and having dealt
> > with X configuration details for many years I would agree.  Running all
> > the X apps via a compatibility mode should not be a problem because it
> > is the same X display code just connecting to the new Mir system which
> > is driving the hardware.  Mac users know that this approach works well.
> > 
> > The recent controversy has been the split between Mir and Wayland, and
> > you can get a flavour of it by looking at some recent Slashdot stories:
> > 
> > http://slashdot.org/index2.pl?fhfilter=wayland
> 
> I've looked at a couple of these slashdot stories and I'm afraid they
> have amplified my concerns. Some of the issues with X11 that are cited
> by Wayland or Mir evangelists are understandable and a few seem
> justified to me. However, their approach to address or resolve these is
> based on abstractions and concepts that are problematic. They appear
> rather narrowly informed by use cases from multimedia and mobile
> computing contexts and seem keenly aware of problems specific to these
> contexts (such as difficulties preventing artifacts in animations),
> and to be rather unaware of the value of many X11 features and virtues,
> such as server-provided fonts (Wayland leaves that to the clients as
> far as I understand).
> 
> For scientific applications this does not bode too well, e.g. the
> occasional flicker upon resizing an R window showing a boxplot is
> no big deal for anyone, but having to draw the lines of the barplot
> pixel by pixel, without server support for rendering lines and text,
> is not going to be nice for whoever gets to write a new graphics
> device for R.
> 
> > The initial effect of the Mir announcement seems to have been a spurring
> > on of the Wayland developers to get the system working.  If Mir and
> > Wayland both become viable solutions then this is no bad thing!
> > 
> > Regarding Ubuntu 14.04, which should form the basis of Bio-Linux 8, I
> > don't think there is a need for concern.  If Mir with X compatibility
> > works well then we can use it.  If not, then others in the community
> > will ensure that X server packages continue to be available because
> > many, many users want these things, and I will build Bio-Linux with
> > these packages.
> > 
> > One significant difference between Mir and Wayland is that Wayland aims
> > for excellent remote desktop/remote application support while Mir
> > currently does not.  Remote X is essential for many Bio-Linux users so
> > I'll do whatever is needed to retain this feature.
> 
> I'm not too sure about Wayland remoting, according to Wikipedia
> (http://en.wikipedia.org/wiki/Wayland_%28display_server_protocol%29)
> it doesn't have network transparency currently, and the pixmap
> centric concept of Wayland compares unfavourably to the vector
> oriented primitives that X11 servers provide for lines etc., and
> this is in fact one of my key concerns about Wayland's use of
> pixmaps, rather than graphics primitives, as the core abstraction.
> 
> > Regarding dropping APT/Dpkg, there's no way that Ubuntu could do this
> > for the 14.04 release and I've seen no sign of them reducing support for
> > things like the Launchpad PPA service.  If Ubuntu did drop Debian
> > packaging then we'd almost certainly switch Bio-Linux back to Debian,
> > but I don't think this is any concern right now.  I think you just have
> > to regard Ubuntu Touch as a separate Ubuntu-branded Canonical product,
> > very much separate from the Ubuntu OS which is still being positioned as
> > a server/workstation solution.
> 
> Ok, that's a reassuring perspective, thanks.
> 
> Best regards, Jan
> 
> > Cheers,
> > 
> > TIM
> > 
> > On Mon, 2013-07-01 at 11:57 +0100, jan kim via RT wrote:
> > 
> > >      Subject: Ubuntu Mir Concerns
> > >        Owner: Nobody
> > 
> > > 
> > > Dear Tim
> > > 
> > > I read that Ubuntu plans to dump X11:
> > > 
> > >     http://fridge.ubuntu.com/2013/06/27/mir-plans-in-13-10/
> > > 
> > > and I'm concerned that this may result in major disruptions of usability
> > > of Ubuntu for bioinformatics and scientific computing in general. No xterm,
> > > no x11 device for R, no recognition of xresources in general, no x11
> > > forwarding, no fvwm2, ...?
> > > 
> > > As far as I understand there is some emulation mode or similar to
> > > smoothen the transition, but I wonder how well this would work, and
> > > whether it will be supported in the long term.
> > > 
> > > I've seen a demo of "Ubuntu Touch" on a tablet a couple of weeks ago,
> > > and was very disappointed to learn that that didn't use the "normal"
> > > Debian package management system, instead Canonical has apparently
> > > developed something that is like those app markets operated by Apple
> > > and Google -- so there was no way to install R, LaTeX, emacs etc. on
> > > a tablet running Touch, as I had hoped, and from this perspective I'm
> > > concerned that some tension may develop between the direction Canonical
> > > seems to pursue with Mir and Touch and the use of Linux for scientific
> > > computing.
> > > 
> > > Do you have any thoughts on this, and how Biolinux will be affected?
> > > 
> > > Best regards, Jan
> > > --
> > > Jan T Kim
> > > Head of Bioinformatics
> > > The Pirbright Institute (formerly IAH)
> > > Ash Road, Pirbright, Surrey GU24 0NF, United Kingdom (GB)
> > > t: +44 1483 231070
> > > e: jan.kim at pirbright.ac.uk (or jttkim at gmail.com)
> > > w: http://www.research.pirbright.ac.uk/bioinformatics
> > >    http://www.pirbright.ac.uk/
> > > 
> > > 
> > > 
> > > ________________________________
> > > 
> > > The information contained in this message may be confidential or legally privileged and is intended solely for the addressee. If you have received this message in error please delete it & notify the originator immediately.
> > > Unauthorised use, disclosure, copying or alteration of this message is forbidden & may be unlawful.
> > > The contents of this e-mail are the views of the sender and do not necessarily represent the views of the Institute.
> > > This email and associated attachments has been checked locally for viruses but we can accept no responsibility once it has left our systems.
> > > Communications on Institute computers are monitored to secure the effective operation of the systems and for other lawful purposes.
> > > 
> > > The Pirbright Institute is a company limited by guarantee, registered in England no. 559784.
> > > The Institute is also a registered charity.
> > > 
> > > 
> > > 
> > 
> > -- 
> > Tim Booth <tbooth at ceh.ac.uk>
> > NERC Environmental Bioinformatics Centre 
> > 
> > Centre for Ecology and Hydrology
> > Maclean Bldg, Benson Lane
> > Crowmarsh Gifford
> > Wallingford, England
> > OX10 8BB 
> > 
> > http://nebc.nerc.ac.uk
> > +44 1491 69 2705
> > _______________________________________________
> > Bio-Linux-dev mailing list
> > Bio-Linux-dev at nebclists.nerc.ac.uk
> > http://nebclists.nerc.ac.uk/mailman/listinfo/bio-linux-dev
> 

-- 
Tim Booth <tbooth at ceh.ac.uk>
NERC Environmental Bioinformatics Centre 

Centre for Ecology and Hydrology
Maclean Bldg, Benson Lane
Crowmarsh Gifford
Wallingford, England
OX10 8BB 

http://nebc.nerc.ac.uk
+44 1491 69 2705