DATALAD-NO-ANNEX(1)

🌸 datalad no-annex 🌸

okay bestie, quick vibe check first: this page lives in the datalad manual, section 1. that's the whole header situation, we're moving on 💅

---

✨ NAME ✨

datalad no-annex — sets up a dataset so some content, like, never ever ends up in that dataset's annex. periodt. it literally tells the annex "not you, girl" 🚫📦

---

🦋 SYNOPSIS 🦋

datalad no-annex [-h] [-d DATASET] [--pattern PATTERN [PATTERN ...]]
       [--ref-dir REF_DIR] [--makedirs] [--version]

look at her go ✨ our main character energy command 🎀

---

💖 DESCRIPTION 💖

so this is so useful for mixed datasets — the ones that also have textual data in them, like source code. that kind of stuff is genuinely better off managed directly in Git, where it's efficient and way more convenient. no cap 🧚

the patterns generally look like this:

code/*

which would match all files in the code directory. slay 💅

but if you wanna match all the files under code/, including everything in all its subdirectories, you use a pattern like this instead:

code/**

that ** is doing the most and we respect it 🔥

lowkey important: this command works incrementally. so any existing configuration (like from a previous plugin run) gets amended, not replaced. it does not tear down what's already there, it just adds. remember that fr fr 💀

---

👑 OPTIONS 👑

• -h, --help, --help-np — shows you this help message, bestie. and --help-np forcefully disables the use of a pager when it displays the help message. no pager, no scrolling drama 🙈
• -d DATASET, --dataset DATASET — specify the dataset you wanna configure. if no dataset is given, an attempt is made to identify the dataset based on the current working directory (it'll try to figure it out itself, icon behaviour ✨). constraints, and yes they're real: value must be a Dataset or a valid identifier of a Dataset (e.g. a path), or value must be NONE.
• --pattern PATTERN [PATTERN ...] — the list of path patterns 🎀 any content whose path matches any of the patterns will not be annexed when it's added to a dataset — instead it gets tracked directly in Git. note that path patterns have to be relative to the directory given by the --ref-dir option. by default, patterns should be relative to the root of the dataset. iykyk.
• --ref-dir REF_DIR — the relative path (within the dataset) to the directory that's gonna be configured. all patterns are interpreted relative to this path, and the configuration gets written to a .gitattributes file in this directory. default: '.' 🫶
• --makedirs — if you set this, any missing directories will be created, so that a file can actually be placed into --ref-dir. it's giving "build the folder so the folder can exist" 💀✨
• --version — shows the module and its version which provides the command. cute little provenance moment ⭐

---

🎀 AUTHORS 🎀

datalad is developed by The DataLad Team and Contributors <team@datalad.org>. literal icons, all of them 👑💜

---

and that's the page!! this is datalad no-annex version 1.1.5, dated 2025-06-15 — go forth and keep your source code out of the annex, bestie 🌈🦋🍓