diff options
| author | Clément Pit-Claudel | 2020-03-19 18:42:10 -0400 |
|---|---|---|
| committer | Clément Pit-Claudel | 2020-03-19 18:42:10 -0400 |
| commit | 47d92a69773755e2ad5d5f987f87337fdf7e98d8 (patch) | |
| tree | 3a074b1269952ef2cbbb4d5fce64531580c61443 /doc/sphinx/proofs | |
| parent | 9f680f776140c8b3b8f79013262d5bd73761d571 (diff) | |
| parent | 1be31dea4cfd31522898edc07fee0829fea7c68d (diff) | |
Merge PR #11601: [refman] Move chapters into new structure.
Reviewed-by: jfehrle
Diffstat (limited to 'doc/sphinx/proofs')
| -rw-r--r-- | doc/sphinx/proofs/automatic-tactics/index.rst | 20 | ||||
| -rw-r--r-- | doc/sphinx/proofs/creating-tactics/index.rst | 34 | ||||
| -rw-r--r-- | doc/sphinx/proofs/writing-proofs/index.rst | 34 |
3 files changed, 88 insertions, 0 deletions
diff --git a/doc/sphinx/proofs/automatic-tactics/index.rst b/doc/sphinx/proofs/automatic-tactics/index.rst new file mode 100644 index 0000000000..a219770c69 --- /dev/null +++ b/doc/sphinx/proofs/automatic-tactics/index.rst @@ -0,0 +1,20 @@ +.. _automatic-tactics: + +===================================================== +Built-in decision procedures and programmable tactics +===================================================== + +Some tactics are largely automated and are able to solve complex +goals. This chapter presents both some decision procedures that can +be used to solve some specific categories of goals, and some +programmable tactics, that the user can instrument to handle some +complex goals in new domains. + +.. toctree:: + :maxdepth: 1 + + ../../addendum/omega + ../../addendum/micromega + ../../addendum/ring + ../../addendum/nsatz + ../../addendum/generalized-rewriting diff --git a/doc/sphinx/proofs/creating-tactics/index.rst b/doc/sphinx/proofs/creating-tactics/index.rst new file mode 100644 index 0000000000..1af1b0b726 --- /dev/null +++ b/doc/sphinx/proofs/creating-tactics/index.rst @@ -0,0 +1,34 @@ +.. _writing-tactics: + +==================== +Creating new tactics +==================== + +The languages presented in this chapter allow one to build complex +tactics by combining existing ones with constructs such as +conditionals and looping. While :ref:`Ltac <ltac>` was initially +thought of as a language for doing some basic combinations, it has +been used successfully to build highly complex tactics as well, but +this has also highlighted its limits and fragility. The experimental +language :ref:`Ltac2 <ltac2>` is a typed and more principled variant +which is more adapted to building complex tactics. + +There are other solutions beyond these two tactic languages to write +new tactics: + +- `Mtac2 <https://github.com/Mtac2/Mtac2>`_ is an external plugin + which provides another typed tactic language. While Ltac2 belongs + to the ML language family, Mtac2 reuses the language of Coq itself + as the language to build Coq tactics. + +- The most traditional way of building new complex tactics is to write + a Coq plugin in OCaml. Beware that this also requires much more + effort and commitment. A tutorial for writing Coq plugins is + available in the Coq repository in `doc/plugin_tutorial + <https://github.com/coq/coq/tree/master/doc/plugin_tutorial>`_. + +.. toctree:: + :maxdepth: 1 + + ../../proof-engine/ltac + ../../proof-engine/ltac2 diff --git a/doc/sphinx/proofs/writing-proofs/index.rst b/doc/sphinx/proofs/writing-proofs/index.rst new file mode 100644 index 0000000000..a279a5957f --- /dev/null +++ b/doc/sphinx/proofs/writing-proofs/index.rst @@ -0,0 +1,34 @@ +.. _writing-proofs: + +============== +Writing proofs +============== + +Coq is an interactive theorem prover, or proof assistant, which means +that proofs can be constructed interactively through a dialog between +the user and the assistant. The building blocks for this dialog are +tactics which the user will use to represent steps in the proof of a +theorem. + +Incomplete proofs have one or more open (unproven) sub-goals. Each +goal has its own context (a set of assumptions that can be used to +prove the goal). Tactics can transform goals and contexts. +Internally, the incomplete proof is represented as a partial proof +term, with holes for the unproven sub-goals. + +When a proof is complete, the user leaves the proof mode and defers +the verification of the resulting proof term to the :ref:`kernel +<core-language>`. + +This chapter is divided in several parts, describing the basic ideas +of the proof mode (during which tactics can be used), and several +flavors of tactics, including the SSReflect proof language. + +.. toctree:: + :maxdepth: 1 + + ../../proof-engine/proof-handling + ../../proof-engine/tactics + ../../proof-engine/ssreflect-proof-language + ../../proof-engine/detailed-tactic-examples + ../../user-extensions/proof-schemes |
