Files
blog/content/posts/2019-02-04-fosdem.md
T
2019-02-12 22:06:16 +01:00

169 lines
8.1 KiB
Markdown

---
title: First impressions of FOSDEM
date: 2019-02-04
author: James McDonald
type: post
categories:
- Tech
draft: true
---
It's big. Really big. Just looking at the timetable is daunting, let alone choosing
which talks to attend.
People come to FOSDEM from everywhere. I don't think I've ever heard such a
wide selection of languages at the same place. It really brings home just how
massive FOSS is, and how many people care about it.
# Venue
I've never been to Brussels before. If there's one regret about my FOSDEM trip,
it's that I didn't have more time to spend in this beautiful city which appears
to be full of fun, friendly people. I would definitely recommend visiting with
some time to spare; I certainly will be again.
Also, the beer. Oh my, the beer.
FOSDEM itself is hosted by [Université Libre de Bruxelles
(ULB)](http://www.ulb.ac.be/). The event takes up a [fair
chunk](https://fosdem.org/2019/schedule/rooms/) of the campus. This was again
quite daunting, but the organisers took a lot of effort to make it easy to
navigate from place to place with a lot of clear signage and timetable
information.
PGDay (see below) was at the Marriott Grand Place, which was pretty much a
regular conference venue as far as the event was concerned. Free running coffee
and a chair is all I require. I can recommend their steak, though.
# Stands
There were many stands run by all the FOSS projects we know and love. The
stands were an odd experience compared to a "normal" conference.
Generally at conferences you have a few big corporations and a bunch of smaller
companies. Everyone wants to sell you something you've never heard of and get
your details so they can send you <s>spam</s>marketing information.
Not at FOSDEM. Most of the time I found myself going up to these fantastic
projects like Debian, GitLab or Apache and the conversation was mostly limited
to "you guys are amazing, thanks for doing what you do". It was a good chance
to find out what teams were working on in the FOSS world. In some cases, like
with Matrix for me, it was useful to be able to go and chat to them after
seeing a talk about their project to get some more detail. The stands also
provided the opportunity to donate to projects by buying a wide array of
merchandise.
I learned about a few projects I hadn't heard of too, like
[PostGraphile](https://www.graphile.org/postgraphile/), an tool to create a
GraphQL API pointing at your existing PostgreSQL database;
[IsardVDI](https://isardvdi.readthedocs.io/en/latest/), a really neat tool to
manage KVM virtual desktops targeted at classroom-type environments; and
[Bazel](https://bazel.build/), a build tool to help you only run the tests you
need to.
# Talks
I managed to select a good few talks in the end. Sadly, some of them (looking
at you, [HTTP/3](https://fosdem.org/2019/schedule/event/http3/)) were so
popular they weren't possible to get in to. Luckily, all of them are recorded
and available to view on [the FOSDEM site](https://fosdem.org/2019/).
Here's a sample of those I attended. I recommend you follow the links and check
out the videos yourself if you find them interesting.
## [Java 4..12, Kotlin, Code Coverage and their best friend — bytecode: scandals, intrigues, investigations](https://fosdem.org/2019/schedule/event/kotlin_code_coverage_bytecode/) --- Evgeny Mandrikov
A talk I really enjoyed that I selected largely at random. Evgeny Mandrikov,
works on [JaCoCo](https://www.jacoco.org/jacoco/). They build code coverage
information by looking inside Java class files which allows them to work with
multiple JVM languages. Unfortunately, it turns out that `javac` outputs code
that... isn't quite what you'd expect. It can also be extremely different for
the same code depending on the target JVM version. The talk had a series of
interesting examples of compiler output bytecode demonstrating these issues and
proving that when you're reverse engineering, sometimes you just have to make a
guess.
## [What's new in PostgreSQL 11](https://fosdem.org/2019/schedule/event/whats_new_in_postgresql_11/) --- Magnus Hagander
My colleague Magnus presented what's new and magical in PostgreSQL 11. Some of the ones that interested me were:
* RANGE BETWEEN, allowing you to operate on windows in your data in ranges based on any sortable data type
* Stored procedures, making it much easier to migrate from... other database systems which one might encounter when you need to embed transactions inside your code
* Logical replication of TRUNCATE, which was confusing by its absence
And of course general improvements in performance, with better use of parallelism.
If you're a PostgreSQL user I'd suggest checking the talk out yourself to get all the details.
## [Netflix and FreeBSD](https://fosdem.org/2019/schedule/event/netflix_freebsd/) --- Jonathan Looney
This was a slightly disappointing talk. I was hoping for some information about
the advantages of FreeBSD for this sort of work, but the talk focussed a lot on
the advantages of working closely with upstream FreeBSD and of publishing
Netflix's improvements. That's very true, but it's an odd pitch to a room full
of people at an open source conference.
There were some interesting details about the hardware Netflix use in their CDN
caches and the areas they have to focus on for performance.
## [Matrix in the French State](https://fosdem.org/2019/schedule/event/matrix_french_state/) --- Matthew Hodgson
The French government were looking for a new communication tool and landed on
Matrix. This talk covered some of the design and deployment of this solution,
and the challenges involved. He also talked about the feedback into the open
source project and the improvements we can expect or access now.
I found this talk enthralling, because I had reviewed Matrix for use as a
corporate communications tool a couple of years ago when it was much more basic,
and it was delightful to see how far they have come. Developments like easily
verifiable end-to-end encryption in this sort of tool just sound marvellous.
The improvements they have made in the development version of Riot bring it up
to par with (dare I say better than?) the major players in this field.
Definitely a project to keep an eye on.
## [Introduction to the BASIC Engine](https://fosdem.org/2019/schedule/event/basic_engine/) --- Ulrich Hecht
Given that a Retrocomputing track existed, it would be rude not to attend. This
talk about building a sub-€10 computer to run BASIC sounded like fun. And
indeed it was! Ulrich Hecht talked about his motivations for building such a
computer, the hardware he selected and what's involved in building the machine.
He was specifically focussed on a machine that had good graphics and sound
capabilities, which he demonstrated. It reminded me a lot of programming for
the C64 or Amiga (the BASIC Engine has a blitter).
## [2019 - Fifty years of Unix and Linux advances](https://fosdem.org/2019/schedule/event/keynote_fifty_years_unix/) --- Jon 'maddog' Hall
It was fantastic to have an opportunity to see Jon 'maddog' Hall talk in
person. He was engaging and enthusiastic as he talked about his experiences
living through the entire history of Unix and Linux. He ran out of time in the
end, but luckily a riot was prevented when he was allowed to finish.
There were various events and people marked with asterisks in the presentation
to denote that given time in the pub, more stories would be available. I would
dearly have enjoyed to hear those stories!
I really can't capture his particular charm and humour here, so I strongly
suggest you check out the talk yourself.
# PGDay
I also attended the FOSDEM PGDay the day before the main event. This was a
great event in its own right, with several interesting talks.
I was particularly interested by a couple of the talks.
## Anonymisation
## Encryption
Also I got a fluffy Slonik.
# Thanks
I'd like to thank the fantastic people who take the time and effort to organise
this event. Having been marginally involved in running conventions, I've seen
how hard it can be to wrangle these things. And this event is *ridiculously
huge* --- the statistics in the closing talk had us eating 200kg of waffles
*alone*! The effort involved must be herculean. So, yeah, thanks!