When I say “alphabetical order”, I mean “alphabetical order”
What behavior is being debated
- Thread centers on GUI file managers that use “natural” (a.k.a. numeric‑aware) sorting for filenames, while terminals and classic tools use strict character‑by‑character (lexicographic) sorting.
- Example:
file-2.txtafterfile-10.txtin strict sort vs. afterfile-1.txtin natural sort. - The article’s concrete case: mixed camera filename formats like
IMG_20250820_055436307.jpgvsIMG_20250820_095716_607.jpgget reordered unexpectedly when numeric segments are interpreted as numbers.
Natural vs. strict lexicographic / “alphabetical”
- Many commenters say natural sort is overwhelmingly what users expect: numbered items, versioned files, episodes, screenshots, etc. should order 1,2,3,…,10, not 1,10,2….
- Others argue the opposite: filenames are strings, so sort should be purely lexicographic; “numbers aren’t in the alphabet” and any numeric semantics are extra “magic.”
- Several people note the author is really asking for “lexicographic” or “ASCII/UTF sort”, not “alphabetical,” and that “alphabetical order” is itself underspecified.
Labeling and configurability
- Strong view that the real bug is UI wording and lack of options: if the menu says “Name,” not “Alphabetical,” it’s not lying—just vague.
- Many advocate two modes: e.g., “Name (natural)” vs “Name (strict)” or a deep preference toggle. KDE/Dolphin and Windows (via registry / Group Policy) are cited as offering such switches.
- Others push back: every extra option multiplies complexity and test surface; defaults should favor the majority use case.
Unicode, locales, and standards
- Multiple comments emphasize that “real alphabetical” is messy in Unicode: locale‑dependent ordering, accents, digraphs, case, and digit characters make any single rule arbitrary.
- References to the Unicode Collation Algorithm, CLDR, ICU, and specific tricky locales (e.g., Czech “ch”, Swedish å/ä/ö, Turkish dotted/undotted i).
- Numeric collation (“kn” option, natural sort) is explicitly a configurable extension in these standards, not a universal default.
Edge cases where natural sort is bad
- Hash‑like filenames, GUIDs, random IDs, or hex strings become harder to scan when digit runs are reinterpreted as numbers.
- Mixed date formats, decimals, scientific notation, locale‑specific separators, and leading zeros produce ambiguous expectations; natural sort only “just works” for a narrow subset.
- Users dealing with large photo collections often prefer sorting by metadata (EXIF date) rather than filename at all.
Philosophy: smart defaults vs. dumb tools
- One camp prefers “dumb but predictable” tools that never guess, even if that requires zero‑padding names.
- Another camp prefers “mind‑reading” behavior that fits lay expectations 99% of the time, accepting occasional surprises.
- Several note a broader trend: UIs removing advanced options in favor of opinionated defaults, frustrating power users who want explicit control.