Guides
The reasoning behind the tools
Each tool answers one question quickly. These guides answer the ones that need more than a widget: which option is right, why the obvious choice is sometimes wrong, and what actually goes wrong in production. Written against the same engines the tools run, so the examples are reproducible rather than illustrative.
35 guides
How these guides are written
Every claim here was checked against the implementation the corresponding tool uses — the same function that runs in your browser. When a guide says a decoder rejects a particular input, that is because the decoder on this site rejects it; when it gives a byte count, the count comes from the same encoding call the tool makes. See the methodology for how the code is tested, and who writes this.
The guides are deliberately longer than the tool pages. A tool page is a reference you land on for twenty seconds; a guide is what you read when the twenty seconds did not answer the question.
What each guide contains
The guides share a shape, and it is not the shape of a product page. Each one opens with the problem as it appears in practice — a file that opens correctly in one tool and garbles in another, a token that decodes but should not be trusted, a string comparison that fails on text that looks identical. Then it gives the rule behind the behaviour, with the standard or specification named, and works through an example that can be reproduced with the linked tool.
The last section of each guide is the part most references omit: where the rule breaks. Encodings have characters they do not cover; formats have features they cannot represent; round-trips are not always the identity. A guide that only told you the happy path would be a restatement of the tool's own page, and it would leave you without an answer the first time an input did something unexpected. Naming the failure modes is the point.
Where to start
- If a string will not match. Start with the invisible characters, then NFC and NFD. Between them they explain most “the text is identical but the comparison fails” cases.
- If you are moving data between systems. Base64 versus percent-encoding and CSV quoting cover the two formats that corrupt data most quietly; the config-format guide covers what a conversion loses.
- If you are handling credentials. Hash, encrypt, encode separates the three verbs, and the JWT guide explains why decoding a token proves nothing about it.
- If you are working at the network layer. CIDR subnets and punycode cover addressing and the lookalike domain problem that comes with internationalised names.
Each guide links to the tools it explains, and every tool page that a guide expands on links back to it, so you can move in either direction: from the concept to the widget, or from a result you do not understand to the explanation behind it.