About
MerceMay
A technical reference library for people who build and operate software systems.
What this is
MerceMay publishes explanatory articles about computing. Each piece takes a single technical subject and works through it: the mechanism, the design constraints that produced it, the costs it imposes, and the cases where it fails. The intended reader already writes software and wants to understand what is happening below the interface they use.
The library spans algorithms and data structures, programming languages and their runtimes, operating systems and memory systems, computer architecture and hardware, networking and internet protocols, distributed systems and databases, software engineering and architecture, the web platform, cloud infrastructure, cryptography and security, machine learning and inference systems, and the operational disciplines around all of it.
How it is organised
Every article belongs to exactly one domain and one cluster within that domain. A domain is a broad subject area — databases, compilers, networking. A cluster is a shelf within it — indexes, query processing, storage engines. Clusters hold a sequence of articles ordered from fundamentals toward advanced practice, so reading a cluster front to back builds a coherent picture rather than a pile of disconnected notes.
Clusters are not a strict prerequisite chain. An article on B-tree node layout does not require you to have read the article on relational normalisation, though both live in the same domain. Cross-references between articles point at the background a given piece assumes.
Standards
Technical claims are grounded in primary sources: specifications, standards documents, reference implementations, and vendor documentation. Where behaviour differs between versions or implementations, the article says so rather than presenting one case as universal. Code examples are written to be run and are kept small enough to read in full.
Articles cite the material they depend on. References are listed at the end of each piece and point to the actual document — an RFC number, a paper, a man page, a source file — rather than to a summary of it.
Dates
Each article carries an archive date. It is assigned so the library can be ordered, browsed by month, and syndicated: it places the article in the archive and the feed, and it is stable, so a given article keeps the same date.
Archive dates are not historical publication records. They are not evidence of when a subject was written, and they should not be cited as a publication date. Anything drawn from an article should be attributed to the article itself, which is what the reference list and the source documents at the end of each piece support.
Using the library
- Topics — the full domain and cluster structure, with article counts.
- Archive — every article in reverse chronological order.
- Tags — keyword index for mechanisms and techniques.
- Search — full-text search across all articles.
- RSS — the most recent hundred articles.
Corrections
Errors are corrected in place, and the correction date is shown on the article. If a claim conflicts with a specification or a reproducible experiment, the claim is wrong and will be changed.