Hackfest1403

= Hackfest March 2014 in Karlsruhe =

Post-WSR 2014: 13th to 15th of March 2014 at CEL in Karlsruhe

Attendees

 * Philip Balister
 * Martin Braun
 * Moritz Fischer
 * Attila Kinali
 * Sebastian Koslowski
 * Jan Krämer
 * Sylvain Munaut
 * Marcus Müller
 * Tom Rondeau
 * Balint Seeber
 * Dimitri Stolnikov

GNU Radio Buffer Architecture
Primarily Tom Rondeau


 * Buffers need to be more flexibility for accelerators
 * Modifying buffers is potentially hazardous and not trivial
 * first step: get in-place buffers to work


 * if that works, other buffers should be manageable, too


 * for sync blocks without history, that should be doable


 * for concatenated blocks, the most upstream one needs to know the read pointer position of every single downstream buffers


 * missed something in setting up the buffer readers for all upstream blocks

 fixed afterwards for proof-of-concept


 * interweaved backpressure might lead to inefficiencies


 * forking flowgraphs may need special treatment

QT instrumentation
Tom Rondeau


 * number sink completes the QT toolkit (now equal to WX)
 * nicely configurable, sensible defaults, QT style sheets usable
 * discussion: switch default gen. option to QT (prob. will do)
 * discussion: hide &quot;wrong&quot; gui toolkit for gen. option (problem being that the error message is non-intuitive)
 * Tabs in block properties to reduce complexity

GSoC

 * Discussion of already available Proposals
 * testing of VNA feasibility (Sylvain Munaut)


 * does work theoretically


 * requires patching UHD, pull req. on github


 * data consistent


 * annoying: B210 does not support timed commands, gives no information on time of tuning


 * not encouraging people to generate last minute proposals

GRC
Sebastian Koslowski


 * improved fault handling


 * When block is missing, in the past all connections were undone, the block would just be invisible


 * new: red box, disabled, invalid; gives name of missing block as key


 * all parameter types now string


 * after generation of missing block, everything will fall into place


 * external tool button -&gt; filter design tool


 * external filter tool setting file, but taps just get exported, no external file dependencies of the .grc file


 * tabs


 * new xml element  , as defined in 


 * core affinity, buffer sizes, etc always in Advanced tab


 * If no tabs specified, defaults to General tab (means we don't have to change old XML files)


 * Documentation tab for easier viewing of docs

PFB channelizer
Sylvain Munaut


 * benchmarking
 * trying to improve speed
 * doing a tree of PFBs can sometimes be better than one large PFB
 * channels &gt; 512 can be broken into multiple stages of channelizers
 * cache or other memory constraints may explain why this helps
 * deinterleaving inside the PFB block instead of using a stream_to_streams block didn't work out

QT GUI in external applications
Sylvian Munaut


 * Support for QTGUI sinks in C++ FGs
 * fixing some swig-related bugs

gr-mac
Balint Seeber


 * based on John Malsbury's work
 * revolves around Simple MAC block
 * OFDM as well as GMSK as Radio interface work
 * TUN/TAP and socket to socket
 * implemented ARP cache equivalent (802.3 Tracker)


 * broadcast for unknown MAC addresses


 * demonstration is awesome
 * test application: send an image progressively
 * ARQ actually working
 * Demo'd by bridging Sylvain over a GR link and routed through Balint's laptop to connect with IRC

GPS receiver in GR
Attila Kinali


 * self-built GRC flow graph unfeasible large
 * Literature complicated to the uninitiated
 * GNSS-SDR exists


 * highly complex


 * highly modular


 * framework for receiver generation


 * you feed in receiver parameters


 * builds flow graph on runtime


 * great flexibility, but complicated to understand if you're new to this


 * GNSS seems to be using only one core, not realtime


 * Tom needs to revisit GNSS-SDR, refresh friendship with them

 Corgan already working with GNSS and continuing the relationship

FAQ

 * Discussion on the new FAQ
 * make a meta-FAQ entry point


 * 4 target audiences


 * Students


 * Commercial interested


 * Software Hackers


 * Hams

Some updates to docs and Doxygen
Martin Braun


 * Better looking
 * Search box available
 * Older versions of Doxygen will generate warnings but still complete