| Age | Commit message (Collapse) | Author |
|---|
|
This is intended to be separate from handling of implicit binders.
The remaining uses of declare_manual_implicits satisfy a lot of
assertions, giving the possibility of simplifying the interface in the
future.
Two disabled warnings are added for things that currently pass silently.
Currently only Mtac passes non-maximal implicits to
declare_manual_implicits with the force-usage flag set. When implicit
arguments don't have to be named, should move Mtac over to
set_implicits.
|
|
ie default goal selector !
How to do this:
- change the default value of default goal selector in goal_select.ml
- eval the emacs code in this commit message
- compile Coq and in each erroring file repeatedly run
[C-c f] (my/maybe-fix-buller-error) then [C-c C-b] (proof-process-buffer)
until there are no errors (NB the first [C-c f] has no effect).
You need to watch for 2 cases:
- overly deep proofs where the bullets need to go beyond the list in
my/bullet-stack (6 layers is enough the vast majority of the time
though). The system will give you an error and you need to finish
the lemma manually.
- weird indentation when a bullet starts in the middle of a line and
doesn't end in that line. Just reindent as you like then go to the
next error and continue.
~~~emacs-lisp
(defconst my/bullet-stack (list "-" "+" "*" "--" "++" "**")
"Which bullets should be used, in order.")
(defvar-local my/bullet-count nil
"The value in the car indicates how many goals remain in the
bullet at (length-1), and so on recursively. nil means we
haven't started bulleting the current proof.")
(defvar-local my/last-seen-qed nil)
(defun my/get-maybe-bullet-error ()
"Extract the number of focused goals from the ! selector error message."
(when-let* ((rbuf (get-buffer "*response*"))
(str (with-current-buffer "*response*" (buffer-string)))
(_ (string-match
(rx "Error: Expected a single focused goal but " (group (+ digit)))
str))
(ngoals (string-to-number (match-string 1 str))))
ngoals))
(defun my/bullet-fix-indent ()
"Auto indent until the next Qed/Defined, and update my/last-seen-qed."
;; (insert (format "(* %s -> %s *)\n" my/prev-count my/bullet-count))
(when-let ((qed (save-excursion (search-forward-regexp (rx (or "Defined." "Qed.")) nil t))))
(set-marker my/last-seen-qed qed)
(indent-region (- (point) 1) qed)))
(defun my/nth-bullet (n)
"Get nth bullet, erroring if n >= length my/bullet-stack"
(or (nth n my/bullet-stack)
(error "Too many bullets.")))
(defun my/maybe-fix-bullet-error (&optional arg)
"Main function for porting a file to strict focusing.
Repeatedly process your file in proof general until you get a
focusing error, then run this function. Once there are no more
errors you're done.
Indentation commonly looks bad in the middle of fixing a proof,
but will be fixed unless you start a bullet in the middle of a
line and don't finish it in that line. ie in 'tac1. - tac2.\n
tac3.' tac3 will get indented to align with tac2, but if tac2
finished the bullet the next action will reindent.
This is a stateful process. The state is automatically reset when
you get to the next proof, but if you get an error or take manual
action which breaks the algorithm's expectation you can call with
prefix argument to reset."
(interactive "P")
(unless my/last-seen-qed
(setq my/last-seen-qed (set-marker (make-marker) 0)))
(when (or arg (> (point) my/last-seen-qed))
(setq my/bullet-count nil)
(set-marker my/last-seen-qed 0))
(when-let ((ngoals (my/get-maybe-bullet-error)))
(setq my/prev-count (format "%s %s" ngoals my/bullet-count))
(if (= ngoals 0)
(progn
(while (and my/bullet-count (= (car my/bullet-count) 0))
(pop my/bullet-count))
(insert (concat (my/nth-bullet (- (length my/bullet-count) 1)) " "))
(setq my/bullet-count (cons (- (car my/bullet-count) 1) (cdr my/bullet-count)))
(my/bullet-fix-indent))
(setq my/bullet-count (cons (- ngoals 1) my/bullet-count))
(insert (concat (my/nth-bullet (- (length my/bullet-count) 1)) " "))
(my/bullet-fix-indent))))
(bind-key "C-c f" #'my/maybe-fix-bullet-error coq-mode-map)
~~~
|
|
|
|
Forall_inv to assert that a property that is true for every element of a list is true for every element in the tail of the list. The second, Exists_impl, parallels Forall_impl and proves that if there exists an element in a list that satisfies a given predicate, and the predicate implies another proposition, then there exists an element in the list that satisfies the implied proposition. Both of these proofs fill natural gaps within the List library.
|
|
Previously, hints added without a specified database where implicitly
put in the "core" database, which was discouraged by the user manual
(because of the lack of modularity of this approach).
|
|
* ext_in_map
* map_ext_in_iff
* firstn_skipn_comm
* skipn_firstn_comm
* skipn_O
* skipn_nil
* skipn_cons
* skipn_none
* skipn_all
* skipn_all2
* skipn_app
* seq_ap
* skipn_app
* skipn_length
* firstn_skipn_rev
* firstn_rev
* skipn_rev
* seq_app
All proofs by Anton Trunov.
|
|
|
|
|
|
|
|
|
|
Specifically Exists_dec, Forall_dec, and Forall_Exists_dec.
|
|
|
|
automatically instead
|
|
|
|
This old compatibility hint database can be safely removed
now that coq-contribs do not depend on it anymore.
|
|
Since edb55a94fc5c0473e57f5a61c0c723194c2ff414 landed, compat notations
no longer modify the parser in non-compat-mode, so we can do this
without breaking Ltac parsing. Also update the related test-suite
files.
|
|
|
|
There were three versions of injection:
1. "injection term" without "as" clause:
was leaving hypotheses on the goal in reverse order
2. "injection term as ipat", first version:
was introduction hypotheses using ipat in reverse order without
checking that the number of ipat was the size of the injection
(activated with "Unset Injection L2R Pattern Order")
3. "injection term as ipat", second version:
was introduction hypotheses using ipat in left-to-right order
checking that the number of ipat was the size of the injection
and clearing the injecting term by default if an hypothesis
(activated with "Set Injection L2R Pattern Order", default one from 8.5)
There is now:
4. "injection term" without "as" clause, new version:
introducing the components of the injection in the context in
left-to-right order using default intro-patterns "?"
and clearing the injecting term by default if an hypothesis
(activated with "Set Structural Injection")
The new versions 3. and 4. are the "expected" ones in the sense that
they have the following good properties:
- introduction in the context is in the natural left-to-right order
- "injection" behaves the same with and without "as", always
introducing the hypotheses in the goal what corresponds to the
natural expectation as the changes I made in the proof scripts for
adaptation confirm
- clear the "injection" hypothesis when an hypothesis which is the
natural expectation as the changes I made in the proof scripts for
adaptation confirm
The compatibility can be preserved by "Unset Structural Injection" or
by calling "simple injection".
The flag is currently off.
|
|
This reverts commit 3a2753bedf43a8c7306b1b3fc9cb37aafb78ad7a.
|
|
"exists c1, c2".
|
|
|
|
|
|
|
|
|
|
|
|
The option is still there, but not documented since it is too
dangerous. Hints and type classes instances are not taking cleared
variables into account.
|
|
|
|
|
|
|
|
|
|
|
|
Conflicts:
tactics/eauto.ml4
(merging eauto.ml4 and adapting coq_micromega.ml to new typing.ml API)
|
|
List.nth_error and List.hd_error were the only remaining places in
the whole stdlib to use type "Exc" instead of "option" directly.
So let's simplify things and use option everywhere. In particular,
during teaching sessions about lists, we won't have anymore to explain
the (lack of) difference between Exc,value,error and option,Some,None.
This might cause a few incompatibilities in proof scripts, if they
syntactically expect "value" or "error" to occur, but this should
hopefully be very rare and quite easy to fix.
|
|
|
|
equalities in configurations like
x=y
x=z
===
P(x,y,z)
where it now produces
===
P(z,z,z)
In particular (equations are processed from most ancient to most recent).
Thanks to this, a "repeat subst" can just be a "subst" in List.v.
Incidentally: moved a nf_enter to enter in subst_one, since the latter
is normally called from other tactics having normalized evars.
|
|
|
|
Since [map_ext_in] is more general, no need to have the same proof twice.
|
|
Slightly broader version of the existing [map_ext]: two [map] expressions
are equal if their respective functions agree on all arguments that are
in the list being mapped.
|
|
NB: this is work-in-progress, there is currently only one
provided implementation (MMapWeakList).
In the same spirit as MSets w.r.t FSets, the main difference between
MMaps and former FMaps is the use of a new version of OrderedType
(see Orders.v instead of obsolete OrderedType.v).
We also try to benefit more from recent notions such as Proper.
For most function specifications, the style has changed : we now use
equations over "find" instead of "MapsTo" predicates, whenever possible
(cf. Maps in Compcert for a source of inspiration). Former specs are
now derived in FMapFacts, so this is mostly a matter of taste.
Two changes inspired by the current Maps of OCaml:
- "elements" is now "bindings"
- "map2" is now "merge" (and its function argument also receives a key).
We now use a maximal implicit argument for "empty".
|
|
NB: this is work-in-progress, there is currently only one
provided implementation (MMapWeakList).
In the same spirit as MSets w.r.t FSets, the main difference between
MMaps and former FMaps is the use of a new version of OrderedType
(see Orders.v instead of obsolete OrderedType.v).
We also try to benefit more from recent notions such as Proper.
For most function specifications, the style has changed : we now use
equations over "find" instead of "MapsTo" predicates, whenever possible
(cf. Maps in Compcert for a source of inspiration). Former specs are
now derived in FMapFacts, so this is mostly a matter of taste.
Two changes inspired by the current Maps of OCaml:
- "elements" is now "bindings"
- "map2" is now "merge" (and its function argument also receives a key).
We now use a maximal implicit argument for "empty".
|
|
|
|
|
|
More results on set_remove, in particular explicit the NoDup
pre-condition. Show that NoDup is preserved by other operations.
|
|
|
|
|
|
section variable.
For some reason, the subproofs solved by [auto] had started using [Hfinjective] from the section context.
|
|
|
|
|
|
|
|
|
