| Age | Commit message (Collapse) | Author |
|---|
|
|
|
|
|
When comparing 2 irrelevant universes [u] and [v] we add a "weak
constraint" [UWeak(u,v)] to the UState. Then at minimization time a
weak constraint between unrelated universes where one is flexible
causes them to be unified.
|
|
|
|
|
|
In Reductionops.infer_conv we did not have enough information to
properly try to unify irrelevant universes. This requires changing the
Reduction.universe_compare type a bit.
|
|
Previously [fun x : Ind@{i} => x : Ind@{j}] with Ind some cumulative
inductive would try to generate a constraint [i = j] and use
cumulativity only if this resulted in an inconsistency. This is
confusingly different from the behaviour with [Type] and means
cumulativity can only be used to lift between universes related by
strict inequalities. (This isn't a kernel restriction so there might
be some workaround to send the kernel the right constraints, but
not in a nice way.)
See modified test for more details of what is now possible.
Technical notes:
When universe constraints were inferred by comparing the shape of
terms without reduction, cumulativity was not used and so too-strict
equality constraints were generated. Then in order to use cumulativity
we had to make this comparison fail to fall back to full conversion.
When unifiying 2 instances of a cumulative inductive type, if there
are any Irrelevant universes we try to unify them if they are
flexible.
|
|
#5684)
|
|
|
|
|
|
The original contribution is from Clément Pit-Claudel. I updated
his code and integrated it with the Coq build system. Many improvements
by Paul Steckler (MIT).
This commit adds the infrastructure but no content.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
fiat-crypto/OSX
|
|
|
|
|
|
|
|
More precisely, we check that future goals retrieved in run_tactic
have no given_up goals since given_up goals are supposed to be
produced only by Proofview.given_up and put on the given_up store.
Doing the same for the shelf does not work: there is a situation where
run_tactic ends where the same goal is both in the comb and on the
shelf. This is when calling "clear x" on a goal "x:A |- ?p:B(?q[x])"
when the dependent goal "x:A |- ?q:C" is not on the shelf. Tactic
"clear" creates "|- ?p':B(?q'[])" and "|- ?q':C". The "advance" thing
sees that the new comb is now composed of ?p' and ?q' but ?q' is a
future goal which is later collected on the shelf (which ?q' is also
in the comb).
I tried to remove this redundancy but apparently it is
necessary. There is an example in HoTT (file
Classes/theory/rational.v) which requires this redundancy. I did not
investigate why: the dependent evar is created by ring as part of a
big term.
So, as a conclusion, I kept the redundancy.
|
|
|
|
Tactic-in-term can be called from within a tactic itself. We have to
preserve the preexisting future_goals (if called from pretyping) and
we have to inform of the existence of pending goals, using
future_goals which is the only way to tell it in the absence of being
part of an encapsulating proofview.
This fixes #6313.
Conversely, future goals, created by pretyping, can call ltac:(giveup) or
ltac:(shelve), and this has to be remembered. So, we do it.
|
|
Adding also tclSETSHELF/tclGETSHELF by consistency with
tclSETGOALS/tclGETGOALS.
However, I feel that this is too low-level to be exported as a
"tcl". Doesn't a "tcl" mean that it is supposed to be used by common
tactics? But is it reasonable that a common tactic can change and
modify comb and shelf without passing by a level which e.g. checks
that no goal is lost in the process.
So, I would rather be in favor of removing tclSETGOALS/tclGETGOALS
which are anyway aliases for Comb.get/Comb.set.
Conversely, what is the right expected level of abstraction for
proofview.ml?
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
wish #4129)
|
|
|
|
Following up on #6791, we remove support refolding in reduction.
We also update a test case that was not properly understood, see the
discussion in #6895.
|
|
Today, TACTIC EXTEND generates ad-hoc ML code that registers the tactic
and its parsing rule. Instead, we make it generate a typed AST that is
passed to the parser and a generic tactic execution routine.
PMP has written a small parser that can generate the same typed ASTs
without relying on camlp5, which is overkill for such simple macros.
|
|
|
|
We systematically use Wg_MessageView for both the message panel and each
Query tab; we register all MessageView in a RoutedMessageViews where the
default route (0) is the message panel. Queries from the Query panel pick
a non zero route to have their feedback message delivered to their MessageView
|
|
Since 4eb6d29d1ca7e0cc28d59d19a50adb83f7b30a2a universe binders were
declared twice for all records.
Since 4fcf1fa32ff395d6bd5f6ce4803eee18173c4d36 this causes an
observable error for monomorphic records.
|
|
|
|
We limit fixpoints to Finite inductive types, so that BiFinite
inductives (non-recursive records) are excluded from fixpoint
construction. This is a regression in the sense that e.g. fixpoints
on unit records were allowed before. Primitive records with
eta-conversion are included in the BiFinite types.
Fix deprecation
Fix error message, the inductive type needs to be recursive for fix to work
|
|
|
|
if we cannot coerce one constructor type to the other. By invariant
they have a common supertype
|
