Regular Programming Lars Wikman, Andreas Ekeroot

Andreas is a man of many hobbies. Interviewing for example. But sometimes, you get strange questions from strange people, end up feeling scared, or start lying just a bit. Then, perhaps, you tell the story of a bug. Perhaps we shouldn't work during the winter?
Lars doesn't have interviews. More like sales calls. H§e shares his experiences of how to recruitment, both as part of interviews and as a more straightforward recruiter.
Finally: the secret to everything Lars does.
Links
Percy NilegårdHiring Processes with Gergely Orosz - Oxide and Friends (podcast)The Indiana Jones switchGigcity ElixirLars' conference reportChattanoogaNervesAmazon AuroraRewriting the Technical Interview

Stories about Ecto quickly redeeming itself, and of what it takes to introduce foreign keys.
Some of us are super comfortable referencing the ID. Lars dislikes that Ecto needs to be more complicated because of SQL, but the abstractions do hold.
Also: the biggest reason to ever use a ORM! It can be reallynice to come back to one after a tour of plain SQL-land.
Some people have just been bitten so hard by cowboys.
Links
EctoForeign keysRethinkDBReferential integrityAXALantmännenModelForm in DjangoCowboy and PlugDSLUpsertsFragmentsHaxl - DSL for creating queriesSQLAlchemyets

Fredrik wants to think about long-lived code. Lars is offended, Andreas only a little bit so.
Are there other good software development practices out there? Other than the ones focusing on building something quickly? Practices for building software which lives on and is maintained for much longer than we seem to care to admit? Should we remove dependencies over time? The swamp of dependency management and vendoring is probed, gradually shifting into firmware, the horrors of floating point (proper excuses are made), small language models.
Finally, of course, indecent cups of tea.
Links
LagomReactFlux architectureReduxChangelog episode with Justin Searls about dependencies as liabilitiesKent Beck talking about managing risks in software developmentKent Beck drawing on a whiteboard and staring at the audienceMithril.jsInteract.jsVendoringWorking effectively with legacy code - the book about legacy systemsDelphi 5FlaskDynamic linkingSAMLPOSIXLibcGlibcMuslH.264MicrocodeOxide and friendsCoral TPU:sTensorflow lite286PentiumCUDAROCmQuantizationLLaMA

The Saint Valentine's peak passed without issue. Andreas had time for semlor.
Lars has opinions on semlor, and can imagine many possible improvements. Like having an apple. Or a pizza.
Lars has had a nice influx of work, including hardware work using Nerves. Testing and very hackish hot code reloading are both included.
Finally, some thoughts on Linux audio, and musings about the possibility of creating really nice audio tools for the platform.
Links
Saint ValentineThe strangler fig patternThe strangler patternPhoenixCowboySemlaMudcakeThe Swedish chef making chocolate mooseFinnish fastlagsbulle with jamOne of Lars' blog posts about NervesFrank Hunleth - also hot code reloads the way Lars has doneLars' Stream deck library for ElixirStream deckElgato key lightPulseAudioPipeWireRogue amoeba's audio tools for MacJACKCustom APT repositoryQuotes
The Nordics go all awkward and weirdIn my heart, it was a catastrophyHad time for semlorAn unimpressive pastryIt's less messy to have an appleProfessional nervesBuilding with nervesA reasonable enough abstractionThe Rogue Amoeba for Linux

Andreas tells the story of a old system which suddenly exhibited a new and frightening bug. Lars shares similar experiences of things going wrong in new and novel ways.
When things do go wrong, it is so nice to have supervision trees or other things which allow you to hear about problems, not to mention recover from them.
Also covered are some stories about TCP, networks, and timeouts. And a realization that testing the frameworks upon which you build could have saved some bacon, had it just been done a long time ago.
Links
DjangoModel-view-controllerDrupalUnicode collationSupervision treesOxide and friends - episode 27TCP_NODELAYQUIC and HTTP/3UDPNyqvist-Shannon sampling theoremHexagonal designQuotes
Gaming convention management systemWhen I say view, I mean controllerView is a better wordIf I ignore it, it will go awayDestructive favouritesAlternative class hierarchiesFailed in new and novel waysBoth a mistake, and interestingAaah, circumflex!TCP the good parts

Lars dove into data pipelines, and emerged bearing arrows and wishing for a lot fewer copies.
What is there to think about regarding data pipelines, what is interesting about them?
Which tools are out there, and why might you want to use them?
Why all this talk about making fewer copies of data?
What does Lars' current ideal pipeline look like, and where does Elixir fit in?
Links
Matt TopolApache ArrowLarge language modelsVector searchBigQuerysedAWKjqReplacing Hadoop with bash - "Command-line Tools can be 235x Faster than your Hadoop Cluster"HadoopMapReduceUnix pipesDirected acyclic graphtee - to "materialize inbetween states"Apache BeamApache SparkApache FlinkApache PulsarAirbyte - shoves data between systems using connectorsCronjobFivetran - Airbyte competitorApache AirflowETL - Extract, transform, loadDesigning data-intensive applicationsStream processingEphemeralityData lakeData warehouseThe people's front of JudeaDBT - SQL-SQL batch-work-thingySQL with Jinja templatesSnowflake - data warehouse thingScalaBroadwayOban - "robust job processing for Elixir"Dashbitpandas - Python data libraryAPLArrow flightGRPCDataFusion - query execution enginePolars - "DataFrames in Rust"Explorer - built on top of PolarsVoltron dataThe Composable CodexPyarrow - Arrow bindings for PythonQuotes
I've been reading a lot about data pipelinesWhat's so special about data pipelines?There's a lot of special toolingThere's a lot of bad, bad toolingLess than optimal toolingConverging on something biggerlkHe got me eventuallyAll of your steps in one bucketWhat tools do you associate with data?I inherited a data pipelineBashReduceIterate on the L and the TThe modern data stackAnd then you demand more workNo unnecessary copiesBarely a copyReconnecting with my Python roots