Cranny

Podcast 2 — In conversation

Speaker A: We searched every nook and cranny of the house for that passport — every drawer, every shelf, every coat pocket. Found it eventually in the inside pocket of a jacket I hadn't worn since March.

Speaker B: Every nook and cranny as the idiom of thorough search — the exhaustive investigation of all possible locations. The idiom is particularly effective because it implies not just the obvious places (the nooks, the visible recesses) but the hidden ones too (the crannies, the gaps and crevices). A search of nooks and crannies leaves nowhere unexamined.

Speaker A: The old cottage walls are full of crannies — you can see daylight through several of them. Very atmospheric in the daytime, somewhat less charming when the wind gets up in November.

Speaker B: Crannies in old stone walls as both character and problem — the atmospheric quality of an old wall with its gaps and irregularities is genuine, and so is the practical problem of draughts, moisture, and cold in winter. The cranny is picturesque from the outside and uncomfortable from the inside.

Speaker A: The database search needs to reach into every nook and cranny of the data model — we can't afford to miss records just because they're stored in an unusual table or a legacy schema nobody maintains.

Speaker B: Nook and cranny as a data discovery metaphor — the data equivalent of the thorough search. Legacy systems typically have crannies: obscure tables, rarely-joined schemas, deprecated fields that still contain live data nobody knows about. A data audit that reaches into every cranny is one that maps the full landscape, not just the obvious main tables.

Speaker A: People sometimes confuse cranny and crevice.

Speaker B: A crevice is typically larger and deeper than a cranny — a crack in rock or earth that may be quite significant in scale. You can fall into a crevice; you wouldn't fall into a cranny. A cranny is distinctly small and intimate — a mouse can fit in a cranny; a person cannot. Both describe gaps or cracks in surfaces, but scale distinguishes them: the cranny is the diminutive of the crevice family.

Speaker A: And cranny is different from cavity. A cavity is an enclosed hollow space — a cave, a gap in a tooth, a space within a wall. A cranny is a narrow, open gap rather than an enclosed space. You can see into a cranny from the outside; a cavity is typically hidden within. The cranny opens to the surface; the cavity is interior.

Podcast 3 — Prompt engineering

Speaker B: Scenario one — a data discovery prompt.

Design a data cranny discovery tool for a database audit platform. The tool searches every nook and cranny of a PostgreSQL database for undocumented, orphaned, or overlooked data. It identifies: tables with no foreign key relationships (isolated data islands), columns with no data in them for more than 6 months (abandoned fields that may contain legacy data), tables not queried by any current application based on pg_stat_user_tables (potentially orphaned tables), and columns whose data type is text but whose values follow a consistent structured pattern (crannies where structured data is being stored in unstructured fields). Each finding is flagged for review with the discovery context.

Speaker B: Tables not queried by any current application based on pg_stat are the database crannies where legacy data lurks — nobody knows about them, nobody maintains them, and occasionally they contain something important.

Speaker A: Text columns with structured values are the data model crannies where someone stored a JSON blob or a delimited list rather than creating a proper structure.

Speaker B: Next — a full-text search cranny prompt.

Build a comprehensive search system for a knowledge management platform that finds information in every nook and cranny of the organisation's data. The search: covers documents (Word, PDF, text), emails in the company email archive, meeting notes in the calendar system, code comments and README files in the source repositories, Slack or Teams message archives, and database records in the CRM and project management systems. Results from all sources are unified in a single search interface with source attribution. The search highlights the cranny location — the specific file, channel, or table — where each result was found.

Speaker A: Code comments and README files as indexed search sources is the developer cranny — useful information hidden in inline comments that is invisible to non-developer search tools.

Speaker B: Right. Now an ecological monitoring prompt.

Design a crevice and cranny habitat survey system for a coastal ecological monitoring platform. The system records and analyses invertebrate species found in rock crevices and crannies at intertidal survey sites. For each survey site: map the GPS coordinates of individual crannies of sufficient size, record the species found in each cranny (anemones, blennies, crabs, periwinkles, limpets, starfish), track the microhabitat characteristics of each cranny (aspect, water retention, depth, substrate type), and compare species diversity between exposed surfaces and cranny habitats. A biodiversity index shows how much richer cranny habitats are in species compared to exposed rock.

Speaker B: Biodiversity comparison between exposed surfaces and cranny habitats is the ecological argument for conserving cranny habitats — the species richness per square centimetre of cranny consistently exceeds that of exposed rock at comparable sites.

Speaker B: Fourth — a codebase exploration prompt.

Write a codebase cranny explorer for a legacy code audit tool. The explorer investigates every nook and cranny of a large codebase to find: dead code paths that are never executed (covered by a condition that can never be true), commented-out code blocks that have been left in place for more than 90 days, TODO and FIXME comments sorted by age (oldest unresolved crannies first), functions defined but never called, and environment-specific configuration values hardcoded in source files rather than in environment variables. Each finding is reported with the file path, line number, and the number of days since it was last modified.

Speaker A: TODO comments sorted by age with the oldest unresolved crannies first surfaces the technical debt that has been quietly accumulating in the corners of the codebase — a 3-year-old TODO is not a reminder, it is a decision that needs to be made.

Speaker B: Fifth — a heritage building survey prompt.

Design a heritage building condition survey app for architectural conservators. The app surveys historic stone buildings with a focus on masonry crannies, cracks, and gaps. For each recorded cranny or crack: log GPS coordinates and location description, photograph the cranny with scale reference, classify the crack type (settlement, thermal movement, structural, erosion), measure the width and depth, record any plant growth or biological colonisation in the cranny, and assess the moisture penetration risk. The app generates a condition report showing the distribution of crannies across the building and ranking them by structural risk.

Speaker B: Biological colonisation in the cranny as a recorded variable captures the dual nature of the masonry cranny — a crack that has been colonised by mosses, lichens, or small plants has both a biodiversity value and an accelerating erosion risk as root growth widens the crack.

Speaker B: Sixth — a full small app.

Build a single-page web app called Nook and Cranny — a personal hiding spot and storage log. The user logs all the small, easily forgotten places in their home where they store things: the cranny behind the radiator where extension cables live, the nook behind the couch cushions where TV remotes hide, the drawer cranny where the spare batteries are, the coat pocket cranny with last winter's shopping list. For each entry: location description, what is stored there, and a date last checked. A search box finds where something is stored. A check-up reminder appears for entries not reviewed in 6 months. All data in localStorage.

Speaker A: Check-up reminder for entries not reviewed in 6 months is the practical feature that prevents the cranny log from itself becoming the thing you forget about — a storage log that you stop updating is just a memory of where things used to be.

Speaker B: That's the lesson — the cranny is the smallest version of a gap, and it is often the most interesting. The crannies of an old wall shelter mosses and memories. The crannies of a database hide orphaned tables and forgotten fields. The crannies of a codebase contain three-year-old TODOs and dead code paths. Searching every nook and cranny is not just thorough — it is where the most surprising discoveries are usually made.

Speaker B: Prompt example: