Research on which query language to use when performing search/query operations over HTTP — specifically in the context of the new HTTP QUERY method.
Use this research to figure out, for your own API, whether one of the 68 query languages analyzed here already fits the query content of a QUERY request — or whether none of them do, and you're better off designing a brand new, purpose-built query language instead.
For years, APIs that need to run a search or a query too complex to fit in a URL have had to misuse HTTP POST — a method that is neither safe nor idempotent — just to carry a query in the request body. The HTTP QUERY method closes that gap: it's an HTTP method that, like GET, is safe, idempotent, and cacheable, but like POST, can carry an arbitrary request body (the "query content"). After more than a decade of on-and-off work in the IETF, QUERY is now a published Internet Standard: RFC 10008, The HTTP QUERY Method, published in June 2026.
That request body has to be expressed in some language. QUERY is deliberately media-type agnostic — the body can be SQL, GraphQL, JSONPath, XSLT, a URL-encoded form, or anything else, as long as the server understands it (advertised via the Accept-Query response header defined by the same spec). That raises an obvious follow-up question: out of the many query languages already in use across databases, search engines, and APIs, which ones are actually good fits for the query content of an HTTP QUERY request?
This repository is an attempt to answer that question systematically, by researching and scoring a broad set of existing query languages against a common rubric, so the results can be used as raw material for a future comparison article — not to declare a single "winner," but to give API designers real, sourced data points when they choose (or design) a query language for their own QUERY-based API.
QUERY is defined in RFC 10008, The HTTP QUERY Method, an IETF Proposed Standard authored by Julian Reschke (greenbytes), James M. Snell (Cloudflare), and Mike Bishop (Akamai), with early contributions from Ashok Malhotra. It was published in June 2026 through the IETF's HTTP working group, after several years of development as an Internet-Draft.
In short, the spec defines:
QUERY: a request method that asks a server to run a query (described by the request content) scoped to the target URI, without altering server state. It's safe and idempotent likeGET, but — unlikeGET— the query itself lives in the request body instead of the URL, avoiding URL-length limits and awkward encoding of complex query parameters.Accept-Query: a response header a server can use to advertise which query-content media types (e.g.application/sql,application/jsonpath,application/x-www-form-urlencoded) it supports for a given resource.- Well-defined semantics for caching (the cache key must incorporate the request content), conditional requests, range requests, and redirection, so that intermediaries and clients can treat
QUERYas a first-class, cacheable operation instead of an opaquePOST.
IANA has registered both QUERY (in the HTTP Method Registry) and Accept-Query (in the HTTP Field Name Registry) as part of the RFC's publication.
For accessible background on why this method exists and what problem it solves, see Bruno Pedro's write-up, The HTTP QUERY Method, on The API Changelog — written while the method was still a draft, but still a good introduction to the motivation behind it (it explicitly calls out the open question of which query language(s) an API should accept in the QUERY body). That article is what originally motivated this research project.
The idea behind QUERY predates its final name by roughly a decade:
- 2015 — the
SEARCHprecursor. Julian Reschke, Ashok Malhotra, and James M. Snell authored an early Internet-Draft,draft-snell-search-method, proposing a safe, idempotentSEARCHmethod that could carry a request body — the same basic problemQUERYwould later solve.- WebDAV already had three safe/idempotent methods that carry a request body —
PROPFIND,REPORT, andSEARCH(the latter from RFC 5323) — and reusing one of them was considered, but ultimately rejected: they're tied to a generic, WebDAV-flavored XML media type rather than being media-type agnostic.
- WebDAV already had three safe/idempotent methods that carry a request body —
- 2019 — discussion reopens. The topic was revived by Asbjørn Ulsberg during that year's HTTP Workshop, prompting the HTTP working group to take a fresh, more general look at the problem.
- The method gets renamed to
QUERY. During the specification's development the method name changed fromSEARCHtoQUERY, better reflecting its relationship to a URI's query component and decoupling it from WebDAV associations. - Standards-track development. The proposal became the IETF Internet-Draft
draft-ietf-httpbis-safe-method-w-body, iterated on for several years within the HTTP working group, accumulating implementation experience and working-group consensus. - June 2026 — publication as RFC 10008. The draft completed IETF review, was approved for publication by the IESG, and was published by the RFC Editor as RFC 10008, The HTTP QUERY Method — a Proposed Standard and, with it, an official IANA registration for both the
QUERYmethod and theAccept-Queryheader field.
- Reschke, J., Snell, J. M., & Bishop, M. (2026). The HTTP QUERY method (RFC 10008). IETF. https://www.rfc-editor.org/info/rfc10008/
- Fielding, R., Nottingham, M., & Reschke, J. (Eds.). (2022). HTTP semantics (RFC 9110). IETF. https://www.rfc-editor.org/rfc/rfc9110
- Fielding, R., Nottingham, M., & Reschke, J. (Eds.). (2022). HTTP caching (RFC 9111). IETF. https://www.rfc-editor.org/rfc/rfc9111
- Nottingham, M., & Kamp, P. (2024). Structured field values for HTTP (RFC 9651). IETF. https://www.rfc-editor.org/info/rfc9651
- IETF HTTP Working Group. (n.d.). Query-method [Issue tracker]. GitHub. https://github.com/httpwg/http-extensions/labels/query-method
- Reschke, J., Malhotra, A., & Snell, J. M. (2015). HTTP SEARCH method (Internet-Draft draft-snell-search-method-00). IETF. https://datatracker.ietf.org/doc/draft-snell-search-method/00/
query-languages/list.md(and the machine-readablelist.csv) — the catalog of query languages under research, each linked to its official documentation.query-languages/evaluation.md— the ten criteria used to evaluate every language (Expressiveness, Simplicity, Flexibility, Community and Ecosystem, Extensibility, Transport Compatibility, Standardization, Security, Performance, Orthogonality), including a detailed breakdown of what "Orthogonality" means for a query language.query-languages/evaluations/— one JSON + Markdown file per language with a scored, sourced evaluation against those criteria, plusrubric.md(the scoring rubric) and aggregatesummary.json/summary.mdcomparison tables grouped by category.
Scores are descriptive snapshots meant as raw research material for a future comparison article, not a ranking of "best" query language — the right choice for a given QUERY-based API still depends heavily on its own data model, transport constraints, and audience.
Want to add a query language that isn't in the catalog yet? This repository ships a GitHub Copilot skill, add-query-language, that automates the whole process: cataloging the language in list.csv/list.md, researching and scoring it against the rubric, creating its evaluation folder, and updating the aggregate summary.json/summary.md (including the Media types table and the Observations section).
To use it in VS Code with GitHub Copilot:
- Open this repository in VS Code with the Copilot Chat extension enabled (the skill lives in
.github/skills/and is picked up automatically). - Open the chat view and either type
/add-query-languagefollowed by the language name, or just ask in plain language, e.g. "Add JMESPath to the query languages portfolio." - Copilot will ask any follow-up questions it needs (e.g. the official documentation URL, a short description) before cataloging, researching, scoring, and evaluating the new language, and updating every affected file.
- Review the changes (new/updated rows in
list.csv/list.md, the newquery-languages/evaluations/<slug>/folder, and the updatedsummary.json/summary.md) before opening a pull request.
This work is licensed under the Creative Commons Attribution 4.0 International License.