slcon3 will be held near Frankfurt/Main on 2016-09-(23-25).
The CfP is closed now. See the schedule below for details.
Please note that we will collect a conference fee for the conference room and lunch/drink package. If the preliminary schedule sticks to Saturday being the only day when we book the conference room, we need to expect around 45 EUR conference fee per person.
(19:30) Welcome in the Biergarten “XXL Waldgeist”, Anselm R Garbe
We will gather together and check out the XXL Waldgeist.
Waldgeist Hofheim Schlossstr. 70 65719 Hofheim www.derwaldgeist.de
(10:00-10:05) Welcome, Anselm R Garbe
Anselm will open slcon3 talk session
(10:10-10:40) libzahl – simple bignum arithmetic, Mattias Andre
Software dependent on algorithms inherently perform better the more complex algorithms. Can we despite this compete with GNU MP and still have a simple solution? This is a discussion about libzahl's design and future.
(10:45-11:15) XML damage control, Silvan Jegen
XML is a horrendously abused file format that suffers severely from over-engineering. There arguably is one legitimate use case for (a subset of) XML however: text markup. In this presentation I will show why that is and how to best deal with XML when encountered in such circumstances. To ease the pain for people like me who have to work with XML every day, I will first give an overview of different parsing strategies employed in XML libraries and introduce the most bearable APIs of the ones available. The second part of the presentation is dedicated to the benchmarking of the most sensible XML libraries and to the discussion of possible alternative parsing approaches as well as their implementation.
(11:20-11:40) Stali Pi B+, Manu Raster
Report on readying Stali for the Raspberry Pi B+ (32 bit). It is also a story about monsters and maiden when we take into comparison other piles of code (kernels, distros etc.) and their build systems. Portability conceptions encountered en route are reviewed and ranked from 1 to 3 (depraved, naive and not too bad).
(11:40-12:00) sRGB and colorspace lightening talk, Laslo Hunhold
(12:00-13:00) Lunch break
(13:15-14:10) SCC and QBE for practical compilation, Roberto E. Vargas Caballero
For now decades, we have witnessed a dramatic increase in the complexity of compilers. Mainstream compilers have become huge C++ programs that try to incorporate all the state-of-the-art research. SCC and QBE propose a new approach to compilation where simplicity and predictability prime over reckless efficiency. These new design constraints have security benefits (as advocated by the "boring crypto" movement) and allow bootstrapping new systems more easily.
(14:15-15:15) The Myrddin Programming Language, Ori Bernstein
Myrddin is a programming language that was written for a saner, easier to use programming environment that I find more pleasant than C. The language lives somewhere between C and ML in design, with parametric polymorphism, type inference, closures, and pattern matching. In some ways, it can be thought of as a suckless rust. It also comes with a complete replacement for many standard libraries on a number of popular (and less popular) platforms.
(15:20-15:35) ii-like chatting improvements, Jan Klemkow
Since slcon2 I have made several improvements in my ii-like chat infrastructure. In this talk I will give an overview of my various activities in this area. I will present new features in my UCSPI tool chain, a new modular front end 'lchat' and the idea of runit integration.
(15:40-16:10) text engine, Francesc Hervada-Sala
The right approach to software systems was introduced by the Unix programming environment over forty years ago with a file system that opened access to data across all applications and with a universal interface model based on strings. This approach must be deepened. Let us define "text" as the semantic data structure that results from parsing strings. Imagine that a software system has a "text engine" at its kernel. When you edit a run control file and save it, the text engine parses it and stores its parse tree into the central repository. When you execute an instruction at the shell, the text engine parses the command line and sets all parameters in the repository before executing the command. An application retrieves a configuration parameter by querying the text engine. In such a system parsers for particular formats or languages would not be embedded into particular applications or commands, they would be stand-alone software units with the sole purpose of converting a string into a semantic representation in the text engine's repository. Applications would not have to deal with and not be tied to particular file formats and languages, and the user would be able to choose the format or language to use for any configuration statement. The user would change the name of a file and would not need to adjust all references to it inside scripts and other files, because the reference in the repository (say as inode) would remain unchanged. To experiment with this new approach I am beginning to develop a text engine
(16:10-16:20) Coffee break
(16:20-16:55) Suckless Image Processing, Enric Meinhardt-Llopis
An image is an array of numbers. Since arrays of numbers are natively represented in C, this language is well-suited for image processing algorithms. Moreover, many complex image processing algorithms are pipelines of simple, independent steps. Thus, the UNIX programming environment is an ideal platform for image processing, where several simple programs written in C share information through UNIX pipes. In this talk, we will see many examples of simple image processing utilities (blur, local maxima detection, interpolation, warping, pointwise algebraic expressions, morphological filters, block matching, optical flow, image fusion) and how they are naturally combined to perform advanced imaging techniques; in the selected example, we show a reflection removal method based on the combination of several images. An important issue, with far-reaching philosophical consequences, is the data format written through the pipes. We explain why the farbfeld image format---proposed by the suckless community---is inappropriate in general, because of the need to represent negative and floating-point pixel values.
(17:00-17:20) shared farbfeld, Jan Klemkow
Farbfeld is a good base for general purpose image processing. The tools that have been made so far demonstrate that sophisticated image processing can be achieved using cooperative programs. In this talk I will present a shared memory interface that speeds up the usage of farbfeld tools. Furthermore, I will sketch the architecture of a suckless-style Photoshop-like graphical image editor, and present a PoC implementation together with some benchmarks.
(17:20-17:30) Display servers, Mattias Andre
(17:35-18:00) stali learnings and beehive observation, Anselm R Garbe
Anselm will present his learnings with stali since last slcon2. In particular he will discuss his new goals with stali and reasoning behind why making stali self-bootstrappable is a bad idea. He will also demonstrate stali as platform for observing his beehives.
Official slcon3 talk session end.
(18:10-19:00) suckless.org e.V. Mitgliederversammlung (yearly suckless.org e.V. member meeting)
- Report of the chairs, Anselm R Garbe, Laslo Hunhold - Report of the treasurer, Jan Klemkow - Other topics (members can raise topics until the begin of the Mitgliederversammlung)
(19:30-*) social event
No talk session plans, hack sessions and departure day.
You can arrange your booking at the venue if you like:
Hotel am Rosenberg Wielandstrae 24 65719 Hofheim am Taunus Germany +49 (0) 6192 292-0 firstname.lastname@example.org
When booking a room, use the booking code: slcon or refer to the suckless.org e.V. event.