Sunday, 8 November 2015

MiniDebconf 2015 ARM, Cambridge - ARM, Cambridge 1640 - final wrap-up - thanks to all

Two days worth of sprints - lots of hugely good work done.

Two days worth of open days with talks - lots of interest, feedback, chatting out of earshot in the breakout rooms.

Two days worth of monetary and other input from all the sponsors:  ARM, Codethink, Cosworth, Hewlett Packard Enterprise Collabora  - PRICELESS

Sociableness in pubs and so on overnight

Thanks also to Steve and Jo McIntyre for a houseful of guests and to all on front desk

And also lastly, again to ARM, as in all these posts because of this great venue and for the small army of ARM pass holders who have let us in and out of doors all this time.

MiniDebconf, ARM, Cambridge - ARM 8 November 1600

 Peter Green's talk very clear and well received.

Now for (almost) the last of the day - Steve McIntyre (93sam / Sledge) for his annual presentation on the state of UEFI

MiniDebconf ARM Cambridge 1525 8 November

Good talk on what it means to transfer LARGE data sets in academia and the downfalls of networking.

Peter Green is about to give a talk on Raspbian - these talks filling in for the missing talk mentioned earlier

Raspbian project - What is it, why it was needed, who is behind it etc.

Lots of typing around me and IRC on #debconf-cambridge

Edit methods - ARM Cambridge - 1500

MiniDebconf postings are being typed on my lap on a laptop real time.

I can't remember the consistent format of how I titled them last time :)

Thanks to all for coffee carrying, to Planet Debian for getting the postings up so quickly, to front desk, organisers, facilitators.

Next up - Network Performance Tuning: Lessons from the Coal Face by Christopher J Walker

[And yes, elgriff, I'm breaking all your good advice I heard less than an hour ago]

MiniDebconf ARM CAmbridge 1400 - ARM, Cambridge 8 November

Andy (rattusrattus) on Video team sprint over the last few days with Ivo de Decker and Stefano Rovera

Background/Existing infrastructure/Why change? [DV/Firewire are EOL]
Things to do

1. Replace twinpact video slide capture system with something that can capture HDMI - better resolution,- digital all the way
Using a pre-production prototype at the moment. CCC also hoping to use this. All kit needs to go into one big box with sensible power supplies
2. Replace DVswitch
Better resolution, select sources, record all inputs, record all edits
Gstreamer based transport solution - import/output format agnostic
GST-Switch - possibly dead? VoctoMix looks more promising - CCC are testing this. First use in anger may be December
3. Improve streaming - single stream out. Encode stream on in-room hardware.
Sharing is a thing ... CCC workflow and setup are good  - everyone should share technology. FOSDEM has 25 rooms ...
4 Replace/expand AV equipment 
Workstation PC - SDI capture/video mixing/local record and encode
Laptop for remote UI
2 cameras
Opsis frame grabber
5 BoF Room recording - lots of options to handle differently
1 Operator maximum / preferably automated
Video conferencing camera / microphone setup
Frame grabber for slides / Gobby feed
6 Logistics
Organise equipment by room
Kit must be kept deployable
Quartermastering necessary
Checklists for each box
Set up a lab for development and release

MiniDebconf Cambridge - ARM, Cambridge, 1340

Lightning talks -

David McBride Scalable infrastructure package distribution - even more mad idea than last year for his scalable package distribution round University of Cambridge - npn

Iain Learmonth - Vmdebootstrap sprint report detailing the work from the Sprint on Thursday/Friday. Live-build-ng needs more unit testing, boot testing, documentation. Working now and some images have been built but still in progress

Helen (elgriff) talking about her weird keyboard
Repetittive strain injuries, carpal tunnel injuries.
Extensive computer use / brass instrument playing AND ignoring pain doing the above. Dont do this.
Vertical keyboards - SafeType and Yogitype.
You should care. If it hurts stop.You don't come with a five year warranty

Stefano Rovera (stefanor) - SVN-Git migration. Debian Python migration team did this for hundreds of packages. "Subversion is really, really evil - it's more evil than you thought"

Sledge - Update on ARM ports. Three ports - armel - older hardware. Armhf - ARMv7 hard float ABI. ARM64 / aarch64 - in Jessie - mostly working. Bits and pieces to be done. Over time, requirements have gone up and there are bits and pieces to work on. For the oldest machines in armel there's an amount of  work to do - port may not survive beyond stretch: manpower / people who care is critical.

MiniDebconf Cambridge - ARM, Cambridge 1115 8 November

Just had a really good couple of sessions: one from the DPL including Q&A, then a good overview of reproducible builds and where we are today  from Holger Levsen (h0lger) and Chris Lamb (lamby)

Then a 2 minute silence at 11.00 as is customary in the UK on this date.

Now Lucy Wayland - who says she has a brainful of IEEE754 from a few years ago working on safety-critical systems so she's unburdening by sharing with all of us - everything you wanted to know about floating point but were afraid to ask (and including a primer on signed/unsigned and fixed point as part of the explanation)

Steve Capper's Java presentation is unfortunately not going to take place, so this afternoon's live stream schedule will be different.

Coffee going down well.