| Age | Commit message (Collapse) | Author |
|---|
|
module)
For the moment, there is an Error module in compilers-lib/ocamlbytecomp.cm(x)a
|
|
The Haskell extraction code would allow line-wrapping of the Haskell
type definition, which would lead to unparseable Haskell code when the
linebreak occured just before the type name. In particular, with a term
name of 46 characters or more, the following Coq code:
Definition xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx := tt.
Extraction Language Haskell.
Extraction xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx.
would produce:
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx ::
Unit
xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx =
Tt
which failed to compile with GHC (according to Haskell's indentation
rules, the "Unit" line must be indented to be treated as a continuation
of the previous line).
This patch always forces the type onto a separate line, and ensures that
it is always indented by 2 spaces (just like the body of each definition).
|
|
|
|
vars by _)
|
|
|
|
|
|
Instead of the original hacks (embedding implicits in string msg in MLexn !)
we now use a proper construction MLdummy (Kimplicit (r,i)) to replace the use
of the i-th argument of constant or constructor r when this argument has been
declared as implicit.
A new option Set/Unset Extraction SafeImplicits controls what happens
when some implicits still occur after an extraction : fail in safe mode,
or otherwise produce some code nonetheless. This code is probably buggish
if the implicits are actually used to do anything relevant (match, function
call, etc), but it might also be fine if the implicits are just passed along.
And anyway, this unsafe mode could help figure what's going on.
Note: the MLdummy now expected a kill_reason, just as Tdummy.
These kill_reason are now Ktype, Kprop (formerly Kother) and Kimplicit.
Some minor refactoring on the fly.
|
|
|
|
|
|
Commit 84c2433a introduced the Any type alias as the Haskell extracted
version of MiniML's Tunknown. However, the code to define the Any
type alias was generated conditional on usf.magic. As it turns out,
sometimes Tunknown appears even if usf.magic is false (i.e., even if
MLmagic does not appear anywhere in the AST). This produced Haskell
code that would not compile; e.g.:
% coqtop
Coq < Extraction Language Haskell.
Coq < Extraction Library Datatypes.
The file Datatypes.hs has been created by extraction.
% ghc Datatypes.hs
[1 of 1] Compiling Datatypes ( Datatypes.hs, Datatypes.o )
Datatypes.hs:261:17: Not in scope: type constructor or class `Any'
Datatypes.hs:261:24: Not in scope: type constructor or class `Any'
The fix is straightforward: produce the code that defines the Any type
alias if usf.tunknown is true.
|
|
|
|
|
|
(Fix for bugs #3470 and #3694)
|
|
|
|
When Coq's Haskell extraction needs to use unsafeCoerce, it passes
the -fglasgow-exts option to GHC, but recent versions of GHC warn
against this:
xx.hs:1:16: Warning:
-fglasgow-exts is deprecated: Use individual extensions instead
This patch does as the warning suggests, replacing -fglasgow-exts
with the specific option that the extraction needs (-XMagicHash).
|
|
When Haskell extraction requires magic type coersion, Coq produces
the following code:
unsafeCoerce :: a -> b
#ifdef __GLASGOW_HASKELL__
import qualified GHC.Base
unsafeCoerce = GHC.Base.unsafeCoerce#
#else
-- HUGS
import qualified IOExts
unsafeCoerce = IOExts.unsafeCoerce
#endif
GHC version 7.6.3 does not allow imports after a type declaration,
and produces this error:
xx.hs:20:1: parse error on input `import'
(referring to the first import statement above). This patch moves
the unsafeCoerce type declaration to just after the import statement,
fixing this compile error.
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16806 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16285 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@16071 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15942 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15844 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15804 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
Scheme comments are output on a single line because Ocaml's Format
module which serves as a backend to Pp has an integer, rather than a string
as identation value, so we cannot make it so that each new line in the
comment starts with ";; ".
I've tried something with Pp.ifb but it was hackish at best and had somewhat
strange results.
Known bug: as Pp.std_ppcmds is non-persistent, the comment is actually printed
only once per Extraction command, even if it outputs several files.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15763 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15715 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
grammar.cma
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15384 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
Util only depends on Ocaml stdlib and Utf8 tables.
Generic pretty printing and loc functions are in Pp.
Generic errors are in Errors.
+ Training white-spaces, useless open, prlist copies random erasure.
Too many "open Errors" on the contrary.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15020 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
The MLcase has notably changed:
- No more case_info in it, but only a type annotation
- No more "one branch for one constructor", but rather a sequence
of patterns. Earlier "full" pattern correspond to pattern Pusual.
Patterns Pwild and Prel allow to encode optimized matchs without
hacks as earlier. Other pattern situations aren't used (yet)
by extraction, but only by P.N Tollitte's code.
A MLtuple constructor has been introduced. It isn't used by
the extraction for the moment, but only but P.N. Tollitte's code.
Many pretty-print functions in ocaml.ml and other have been reorganized
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14734 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
A informative inductive type with one constructor C and one informative arg to C
is normally extracted as an identity, with C removed, see for example
the "sig" type. When this new option is set, these singleton types
are left untouch, providing extracted code which is closer to the initial
Coq development.
Feature requested by Wouter Swiestra.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14260 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14257 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14142 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
We simply factorize code that was already existing for Ocaml.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14011 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
This happens for instance when the main component of the fixpoint
block has been provided via Extract Constant
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13889 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
For Haskell, we still try to provide readable indentation,
but we now avoid relying on this indentation for correctness.
Instead, we use layout-independant syntax with { } when
necessary (after "case of" and "let"). It is much safer this
way, even if the syntax gets a bit more cumbersome.
For people allergic to {;}, they can most of the time do a
tr -d "{;}" without changing the meaning of the program.
Be careful nonetheless: since "case of" is now delimited,
some parenthesis that used to be mandatory are now removed.
Note also that the initial "module ... where" is still without
{ }: even when Format goes crazy it doesn't tamper with column 0.
Other modifications:
- Using "Set Printing Width" now affects uniformly the extraction
pretty-printers. You can set a greater value than the default 78
before extracting a program that you know to be "really deep".
- In ocaml (and also a bit in Haskell), we now try to avoid abusing
of 2-char-right-indentation. For instance | is now aligned with
the "m" of match. This way, max_indent will be reached less frequently.
- As soon as a pretty-print box contains an explicit newline,
we set its virtual size to a big number, in order to prevent this
box to be part of some horizontal arrangement.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13870 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13861 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13859 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13770 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
We now keep some type information in the "info" field of constructors
and cases, and compact a match with some default branches (or remove
this match completely) only if this transformation is type-preserving.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13732 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13461 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13417 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
See http://caml.inria.fr/mantis/view.php?id=4940
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13413 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13323 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
In addition to the "| _ -> cst" situation, now we can also
reconstruct a "| e -> f e" final branch. For instance,
this has a tremenduous effect on Compcert/backend/Selection.v.
NB: The "fun" factorisation is almost more general than the "cst"
situation, but not always. Think of A=>A|B=>A, 1st branch will be
recognized as (fun x->x), not (fun x->A). We also add a fine
detection of inductive types with phantom type variables, for which
this optimisation is type-unsafe.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13267 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
And similarly for Haskell: we do not force capitalized/uncapitalized
filenames anymore, but we rather follow the name of the .v file (with
new extensions of course). Ok, this is an incompatible change, but
it is really convenient, some people where actually already doing
some hacks to have this behavior (cf. Compcert).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13260 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
- Add module parameters in the structure of visible_layer,
in order for module params to be part of name clash detection,
avoiding this way a source of potentially wrong code.
- In case of clash, module params are alpha-renamed to something
unique (Foo__XXX where XXX is the number contained in the mbid).
This solves some situations that were unsupported by extraction.
for instance the "Module F (X:T). Module X:=X. ... End F."
- We now check in Coq identifiers the presence of the
extraction-reserved string __. If it is found, we issue a warning
(which might become an error someday).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13240 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
- For Haskell, modules abbreviations and applied functors are expanded.
The only remaining sitation that isn't supported is extracting functors
and applying them after extraction.
- Add a module extraction for Scheme with the same capabilities as for
Haskell (with no Extraction Library, though).
- Nicer extracted module types (use of the mb.mod_type_alg if present)
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13236 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13235 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
Module types are ignored, functors and module ident raise an error
When dealing with simple modules, even nested, the structure
hierarchy is removed, and names are arranged in the following way:
- For the monolithic extraction, we simply use next_ident_away
on short names, as we do when the same name appears in two .v.
- For modular extraction, A.B.t become A__B__t or _A__B__t
depending whether t is a type, a constructor or a constant.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13210 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
For instance:
Extract Inductive nat => int [ "0" "succ" ]
"(fun fO fS n => if n=0 then fO () else fS (n-1))".
See Extraction.v for more details and caveat.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@13025 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
- Many of them were broken, some of them after Pierre B's rework
of mli for ocamldoc, but not only (many bad annotation, many files
with no svn property about Id, etc)
- Useless for those of us that work with git-svn (and a fortiori
in a forthcoming git-only setting)
- Even in svn, they seem to be of little interest
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12972 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
* An inductive constructor Dummy instead of a constant dummy_name
* The Tmp constructor indicates that the corresponding MLlam or
MLletin is extraction-specific and can be reduced if possible
* When inlining a glob (for instance a recursor), we tag some
lambdas as reducible. In (nat_rect Fo Fs n), the head lams of
Fo and Fs are treated this way, in order for the recursive call
inside nat_rect to be correctly pushed as deeper as possible.
* This way, we can stop allowing by default linear beta/let
reduction even under binders (can be activated back via
Set Extraction Flag).
* Btw, fix the strange definition of non_stricts for (x y).
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@12938 85f007b7-540e-0410-9357-904b9bb8a0f7
|
