DEV Community: Christian Himpe

One Minute: SLED - Schemy LISP en DOS

Christian Himpe — Sun, 30 Aug 2026 10:27:01 +0000

SLED is a purely symbolic Scheme-inspired LISP interpreter for DOS or DOSbox.

Operating System: MS-DOS, FreeDOS
Emulation: DOSBox, DOSBox-X, DOSBox-Staging
Memory Segmentation: Real-Mode
Memory Model: Small

Interpreter Features: Tail-Call Optimization, Lexical Scoping, Mark-and-Sweep Garbage Collector, Trampoline Evaluator
Language Features: 8 Special Forms, 21 Builtin Functions, 21 Standard Functions
Heap Size: 12K (cons cells)
Based on: Kilo LISP

Lines of Code: 1100
Programming Language: C89
Compatible Compilers: MS C 6.0A, OpenWatcom V2
Executable Size: 11KiB

License: 0BSD
Website: sled.neocities.org
Reference: sled.neocities.org/#index
Repository: codeberg.org/gramian/sled

(println 'hello _ 'world)

DatAasee 0.9 Released

Christian Himpe — Fri, 08 May 2026 06:18:02 +0000

DatAasee Tech Stack Canvas

Setting: Many distributed data and metadata sources
Goals:
- Centralize metadata
- Interlinked metadata catalog
- Super-index for bibliographic and research data
Features:
- Interact through HTTP API (JSON)
- Search by filter/facet, full-text, ingest-source, DOI
- Custom queries via: SQL, OpenCypher, MQL, GraphQL, Redis
Frontend: Lowdefy (Optional)
Backend: Connect (formerly Benthos)
Data Storage: ArcadeDB (Graph Database)
Infrastructure: Compose (via Docker or Podman)
Deployment: (Public) Container Images
Monitoring: Container Logs
Integrations:
- Protocols: OAI-PMH, S3, GET, DatAasee
- Encodings: XML
- Formats: DataCite, DC, LIDO, MARC, MODS
Exports: DataCite, BibJSON
Security: Privileged endpoints
Testing: check-jsonschema
Development: https://github.com/ulbmuenster/dataasee

DatAasee 0.5 Released!

Christian Himpe — Thu, 13 Nov 2025 15:48:16 +0000

DatAasee - A Metadata-Lake for Libraries - Version 0.5 has been released:

https://github.com/ulbmuenster/dataasee

Using declarative programming all the way down:

A Compose-based web service,
combining the ArcadeDB database,
with a Benthos back-end,
and a Lowdefy front-end.

DatAasee - A Metadata-Lake

Christian Himpe — Tue, 06 May 2025 15:42:54 +0000

DatAasee is an open-source metadata-lake, functioning as a metadata aggregator of distributed data-sources in libraries. It instantiates a data space and provides a FAIR-compliant metadata-layer for repositories. Its data architecture is a data-lake which is its own metadata catalog, hence a metadata-lake, with an underlying graph data model.

Internally, DatAasee is realized as a three tier software architecture, where each tier is encapsulated in its own container. The inner data-tier is a multi-model NoSQL database ArcadeDB. The middle logic-tier is an (HTTP-JSON-REST) API-server and semantic-layer around the database realized via Connect (aka Benthos). The outer presentation-layer uses Lowdefy as a prototype frontend which exclusively uses the API. This multi-container application is deployed and orchestrated via Compose.

For in-depth explanations, see the open-access white paper: "DatAasee -- A Metadata-Lake as Metadata Catalog for a Virtual Data-Lake".

Scheming About Clojure

Christian Himpe — Wed, 13 Nov 2024 20:53:52 +0000

Clojure is a LISP for the Java Virtual Machine (JVM). As a schemer, I wondered if I should give Clojure a go professionally. After all, I enjoy Rich Hickey's talks and even Uncle Bob is a Clojure fan. So I considered strength and weaknesses from my point of view:

Pros

S-Expressions
Makes functional programming easy
Schemy naming with ? and ! suffixes
Integrated testing framework
Platform independence due to JVM
Simple Java interoperability
Clojure map type corresponds to JSON
Web-server abstraction with extensions (Ring)
Dedicated Ubuntu-based Docker image

Cons

Too many core functions
Too many concurrency concepts
Having collection functions and the sequence API is confusing
Keywords feels unnecessary, given symbols
Unwieldy default project structure
Leiningen feels forced upon you
Clojure is not just a single jar (anymore)
No integrated JSON parser

Insight

Clojure seems good enough. It is not flawless and somewhat overloaded, but far, far ahead of Javascript, Python, Go, or Rust. Of course, I would always prefer CHICKEN Scheme for any passion project. But in an environment that already runs databases written in Java, the JVM has street cred, and a large community hints at sustainability, Clojure presents itself as well balanced in novelty and stability. All in all, Clojure seems to be the enterprise Lisp.

References

One Minute: DatAasee

Christian Himpe — Tue, 10 Sep 2024 15:43:52 +0000

DatAasee is a metadata-lake to aggregate and centralize bibliographic and research-data metadata from distributed data and metadata sources to improve discoverability and FAIRness.

Features: REST-like CQRS HTTP-API, Faceted Search, Full-text Search
Interface: OpenAPI, JSON, JSON API, JSON Schema
Query Languages: SQL dialect, Cypher, Gremlin, MQL, GraphQL
Ingest Protocols: OAI-PMH, S3
Ingest Encoding: XML
Ingest Formats: DataCite, DublinCore, MARC, MODS
Deployment: Compose, Docker-Compose, Podman-Compose, K8s (via Kompose)
Components: ArcadeDB, Connect, Lowdefy
License: MIT

Website: https://ulbmuenster.github.io/dataasee
Repository: https://github.com/ulbmuenster/dataasee
Companion Paper: https://arxiv.org/abs/2409.05512

One Minute: DuckDB

Christian Himpe — Tue, 20 Aug 2024 17:16:46 +0000

DuckDB is an open-source, in-memory, column-oriented relational database management system inspired by SQLite.

Data Model: Relational
Interfaces: C, C++, Go, Java, Julia, Node, Python, R, Rust, Swift, WASM, ADBC, ODBC
Query Language: SQL super-set
Features: Console, ACID
Programming Language: C++
License: MIT

Website: https://duckdb.org
Documentation: https://duckdb.org/docs
Repository: https://github.com/duckdb/duckdb
Mastodon: https://mastodon.social/@duckdb
Twitter: https://twitter.com/duckdb

Database of Databases: https://dbdb.io/db/duckdb
DB-Engines: https://db-engines.com/en/system/DuckDB

One Minute: Compose

Christian Himpe — Sat, 30 Mar 2024 18:51:46 +0000

Compose is a single-host container orchestrator for single and multi-container applications.

Specification: https://compose-spec.io
Definition: Compose File (YAML)
Elements:
- Services,
- Network,
- Volumes,
- Configs,
- Secrets
Supports:
- .env files,
- Interpolation
Features:
- Dependency resolution,
- Resource limits (CPU & RAM),
- Remote deploy (SSH)
Compatibility:

My Data Engineering Library

Christian Himpe — Tue, 19 Mar 2024 14:08:20 +0000

This is neither an "Ultimate List of Data Engineering Books" nor the "Data Engineering Must-Read List", but rather my subjective recommendation for a minimal data engineering library; particularly for data engineers coming from software development, as I used to.

General

Reis, Housley: "Fundamentals of Data Engineering"; Oreilly Media, 2022.
- Discusses all aspects of data engineering and interlinks topics through "under currents".

Data Modeling

Hoberman: "Data Modeling Made Simple"; Technics Publication, 2016.
- Must-read for anybody doing data modeling.
Serra: "Deciphering Data Architecture"; Oreilly Media, 2024.
- Description and comparison of major data architectures.

Databases

Forta: "SQL in 10 Minutes"; Pearson, 2010.
- The best minimal SQL guide I found, yet.
Kaufmann, Meier: "SQL and NoSQL Databases"; Springer, 2019.
- Beyond touching all the database basics, the relational (SQL), document (Mongo) and graph model (Cypher) are recurringly contrasted.

Containerization

Öggl, Kofler: "Docker"; Rheinwerk Computing, 2023.
- Touches on all facets of Docker alongside with typical application examples.
Gkatziouras: "A Developer's Essential Guide to Docker Compose"; Packt, 2022.
- In deep description of the Compose orchestrator.

Extra

Densmore: "Data Pipelines Pocket Reference"; Oreilly Media, 2021.
- Good example-driven overview; original source of EtLT (extract, partial transform, load, transform).
Chromatic: "Extreme Programming Pocket Guide"; Oreilly Media, 2003.
- Extreme programming (XP) is the sanest agile method with some good sustainability practices like the 40h week.

One Minute: Lowdefy

Christian Himpe — Sun, 17 Mar 2024 17:03:07 +0000

Lowdefy is a declarative low-code front-end framework for internal tools or CRUD apps defined via YAML or JSON.

Block Types: Input, Display, Container, List
Connections: Redshift, S3, HTTP, ElasticSearch, Google Sheets, Knex, MariaDB, MongoDB, SQL Server, MySQL, PostgreSQL, Redis, SendGrid, SQLite, Stripe
Features: AgGrid, ECharts, Markdown (Display Blocks)
Modes: Development, Production
Written In: Javascript
Layout: Flexgrid
Build On: Next.js
Requires: PNpM
Deploy Via: Docker
License: Apache-2.0
Website: https://lowdefy.com
Documentation: https://docs.lowdefy.com/introduction
Repository: https://github.com/lowdefy/lowdefy
NPM Package: https://www.npmjs.com/package/lowdefy

YAML-based Database Docu

Christian Himpe — Fri, 08 Mar 2024 13:36:15 +0000

I recently came across the modest yasql standard to document database schemas of SQL databases, which I now use to document my rather involved ArcadeDB schema. My favorite yasql features are:

Very readable,
Rich metadata fields,
A definitions object for custom "macros".

Here is a small example of a yasql docu:

database:
  name: "My DB"
  source: "PostgreSQL"
  project: "My Project"
  version: 1.0
  license: "CC-BY"
  authors:
    - "Me"

definitions:
  tiny: 255

tables:
  myTable:
    myCol: "VARCHAR(tiny)"

NoSQL

Generally, NoSQL databases do not use tables to organize data. For example, ArcadeDB - a multi-model DBMS - provides a document model and a graph model. Hence, instead of a tables object, I use documents, vertexes, and edges objects:

documents:
  myDoc:
    myStr: "STRING (mandatory, notnull, readonly)"

vertexes:
  myVtx:
    myNum: "LONG (default 0)"

edges:
  myEdg:
    myLst: "LIST (max 255)"

Alternatives

There are (open) alternatives available which provide the benefit of visualization as entity-relationship diagram. So why not use ...

... DBML ? Because:
- Column settings have to be from a very limited vocabulary.
- C-Style comments // and /* */.
- Insufficients metadata fields (in the project object).
... D2 sql_table ? Because:
- Every table needs a shape: sql_table, meaning a column cannot be named shape?
- The constraint keyword is superfluous since a constraint is already delimited by curly braces { }.
- No metadata fields.

Visualization

You can visualize a yasql-file using PlantUML, which provides a YAML renderer. For instance, you can find the rendering of the yasql sample here

Practically, I embed a plantuml verbatim block wrapping the yasql YAML, in a Markdown document:

```plantuml
@startyaml

# your yasql goes here ...

@endyaml
```

This way, using the VScode PlantUML Plugin I can get live visualizations of my schema.

One Minute: Lightwhale

Christian Himpe — Sun, 28 Jan 2024 22:17:39 +0000

Lightwhale is a lightweight, immutable Linux for hosting Docker containers.

Based On: Buildroot and Docker
Concept: Immutability and Persistence
Advantages: Small attack surface, no redundant apps, no installation
Compatibility: Docker-Compose
Connectivity: SSH
Platform: x86-64
License: GPL-2.0
Website: https://lightwhale.asklandd.dk
Repository: https://bitbucket.org/asklandd/lightwhale