On Thursday 16 December 2010 17:08:53 Greg Freemyer wrote:
On Thu, Dec 16, 2010 at 5:30 AM, Matt Gray
wrote: reasonable enough reason.
on the other hand, so much is revolutionary in this release including a genuinely awesome KDE4 desktop this time around, project Bretzn, opensource broadcom drivers, speedy kernel patch, bluez, and much more, that it is tempting to give this release the 12.0 moniker.
if the multi-touch xinput extensions make it in I would be all for twelve, as that would make a huge boost to Meego/touch interface respins of suse on arm and intel.
Is there a read-me or draft press release that gives at least a basic overview of the above?
My limited understanding is:
awesome KDE4 desktop this time - i thought we said that about KDE 4.4? What's so great about 4.6?
4.4 was basically usable, 4.6 moves the linux desktop to a new level. Or something like that :D
project Bretzn - never heard of it. I doubt users are clamoring for it.
http://news.opensuse.org/2010/10/26/from-the-developer-to-the-user-and-back- announcing-project-bretzn/' Yes, it's more of a developers thing, but quite awesome imho.
opensource broadcom drivers - nice, but my reading is that this is for limited set of new devices. I'd like to know more about the range of supported devices before considering this a killer feature.
speedy kernel patch - most users are not impacted. Those that are can do this in 11.3 (as I understand it) via a boot time script.
Well, more patches have been added compared to the 11.3 2.6.34 kernel, like Nick Piggin's VM scalability patches that decrease the impact of heavy IO on a linux system. This can actually be quite noticable - and depending on what kernel we finally ship (2.6.37 or 2.6.38) we'll have most or all of these patches.
bluez - I've at least heard of this, but don't know what it is.
Bluetooth? Rings a bell? Bluez brings a much improved bluetooth stack.
multi-touch xinput extensions - very limited set of users, why update the whole distro version for this?
it's new and innovative and hot?
=== If KDE 4.6 is really a huge improvement over 4.4 and we don't expect a similar big jump in the 11.6 timeframe, then I see an issue.
We don't.
If we bump to 12.0 because of KDE as the main reason, then we should also bump to 13.0 when GNOME 3.0 is out.
That sounds like a poor way forward.
Yup. Which is why personally, I'm torn about this... the next release, besides GNOME 3.0, might also have Unity if Nelson and others keep hackin' away - also really cool. Then again, did Smeegol's work make it into factory? If that's possible at all with the latest 'rules' from the linux foundation...
My vote is to wait for GNOME 3.0 to be released and bump to 12.0 with the release that includes it.
Well, KDE IS our default desktop. Then again, GNOME 3.0 brings much more changes to the user experience. Difficult, difficult. Which is why I think we should go with 11.4 as that's what we've been communicating for a long time anyway...
=== OTOH, I'd love to see something like comprehensive and fully integrated support of SSDs (including discard support) in a laptop environment be the driver for the 12.0 moniker.
For environments with a single storage device (eg. a laptop) we are so close, a last minute push could get everything into 11.4, but I think it would need to be project priority and get several people involved.
--> current status as of current factory
The kernel support is there to a major extent with 2.6.37. (ext4 and xfs only I believe)
(I'm ignoring btrfs since its still experimental, and VFAT because I hope a Linux laptop would not use VFAT as a primary FS type.).
The userspace scripts for wiper.sh / hdparm are there already for 11.2 and 11.3 releases. Bug fixes for both were rolled out via the update channel a couple months ago. Obviously factory has these.
A new userspace tool to invoke a new 2.6.37 ext4 background discard feature is supposed to be in 11.4 (it may be in this milestone. Part of linux-utils. The util is trivial and basically just makes a ioctl call. The magic is in the 2.6.37 kernel.)
The gpart partitioning tools and yast partitioner module were updated for 11.3 to align newly created partitions on Erase Block boundaries, so that is already in.
All that's missing is read me files, press releases, a cron entry automatically created to invoke the script if needed and possibly some yast partitioner updates to inform the user that a drive is a SSD and ask how they want discard handled. (There are at least 3 options for ext4 based filesystem. Only one should be used as they provide redundant functionality. The choice is driven by multiple factors.)
And most importantly testing.
fyi: For more background of what I'm talking about, see this wiki page I maintain: http://en.opensuse.org/SDB:SSD_discard_%28trim%29_support
Thanks Greg