<

Dániel Vámosi

Making flight booking simple and teaching robots how to draw what they dreamt about

Dániel started as an architect (not software) 13 years ago, then he gradually switched to software development. Since 2016 Dániel has been leading the tech behind flyiin.com and bringing BEAM technologies to the aviation industry.

Past Activities

Dániel Vámosi
Code BEAM Lite Berlin
11 Oct 2019
15.05 - 15.45

Visual time-travelling using BEAM trace dumps

Few features of the BEAM can create as much excitement as its built-in tracing capabilities. It can directly demonstrate the simplicity of the actor concurrency model. However, tracing production nodes can be tricky. There are many existing attempts to do so by using additional tooling (including my own failed attempt) in deciding what and how to trace. I would like to propose a new method/tool to prove how the visualisation of trace dumps can help in writing well-targeted production trace probes.

OBJECTIVES

I would like to cover the following four topics: - Advantages of having full execution data at hand for better visual layout of trace events - The representation of time in BEAM, how to deal with time warps and the ways to reconstruct the past - How LiveView and the reworked tracing BIFs are helping in bootstrapping custom monitoring tools - How the symbolism of temporal logic can help to construct an easy-to-grasp visual notation (briefly)

AUDIENCE

The proposed method gives beginners a helping hand in understanding the concurrency primitives behind their own code or any lib they were depending on. Intermediate users will benefit from the summary of the way time is represented in the BEAM. Proficient users might find the quick peek into the formal approach of reasoning about time interesting.

Dániel Vámosi
Code BEAM Lite Budapest
20 Sep 2019
13.40 - 14.00

Visual time-travelling using BEAM trace dumps

Few features of the BEAM can create as much excitement as its built-in tracing capabilities. It can directly demonstrate the simplicity of the actor concurrency model. However, tracing production nodes can be tricky. There are many existing attempts to do so by using additional tooling (including my own failed attempt) in deciding what and how to trace. I would like to propose a new method/tool to prove how the visualisation of trace dumps can help in writing well-targeted production trace probes.

OBJECTIVES

I would like to cover the following four topics: - Advantages of having full execution data at hand for better visual layout of trace events - The representation of time in BEAM, how to deal with time warps and the ways to reconstruct the past - How LiveView and the reworked tracing BIFs are helping in bootstrapping custom monitoring tools - How the symbolism of temporal logic can help to construct an easy-to-grasp visual notation (briefly)

AUDIENCE

The proposed method gives beginners a helping hand in understanding the concurrency primitives behind their own code or any lib they were depending on. Intermediate users will benefit from the summary of the way time is represented in the BEAM. Proficient users might find the quick peek into the formal approach of reasoning about time interesting.