| Age | Commit message (Collapse) | Author |
|---|
|
|
|
Two issues in one:
- some focused_simpl were called on the wrong locations
- some focused_simpl were done on whole equations
In the two cases, this could be bad if "simpl" goes too far
with respect to what omega expects: later calls to "occurrence"
might fail. This may happen for instance if an atom isn't a variable,
but a let-in (b:=5:Z in the example).
|
|
|
|
|
|
|
|
|
|
|
|
others.
|
|
|
|
This allows to share the test for possible relocalisation done in envars.ml.
|
|
Also standardizing the choice of the default datadir (I don't see why
we should add by default both /usr/local/share/coq and /usr/share/coq
when we know that the installation is in only one of them).
Open question: test for possible relocation of the installed coq
should be done on raw dirname of the executable or on the
standardization of this name wrt symbolic links?
|
|
This allows to centralize in the configuration file the description of
the 3 possible installation layouts (dispatched over directories
shared by multiple application as in unix, self-contained style like
in windows, local non-installation as with option -local).
Also supporting relocalisation when -prefix or -libdir and co is given.
|
|
This goes towards an approach where a local layout can be seen as an
installed layout.
|
|
Constrintern.pf_global returns a global_reference, not a constr,
adapt plugins accordingly, properly registering universes where
necessary.
|
|
Consequence: docdir is always defined, no more "" in external
preferences for manual and stdlib when using coqide in -local mode.
|
|
They were not used for looking for coqide files in the situation when
the effective installation path happens to be exactly the installation
path proposed by default, while relevant files were however (possibly)
installed in these directories.
|
|
installation paths in unix or win32.
There are two layouts (self-contained or unix-like) and we build
absolute paths from them. Under unix, there is a fully relative layout
(when user gives a prefix) and a standard semi-relative layout (where
most file are under /usr/local with the absolute /etc/xdg/coq as an
exception).
I respected the existing semantics that under cygwin, the unix layout
is the default one when prefix is not given, but the self-contained
layout is the default one when a prefix is given.
|
|
There seems to have been several bugs, when -prefix is given:
- Under win32: "" instead of "lib" (presumably introduced in ab442aed8, 2006)
- Under win32: datadir, mandir, docdir should presumably be in
"share", "man", "doc", as given in default
- Under non-win32: coqdoc files should be in latex subdir not emacs subdir
|
|
|
|
"libdir" to "COQLIBINSTALL" then "libdir", then "coqlib".
For the record, here is how installation options are named at the
current time in the different places they are used (if any):
Name in Name in Name in Name of option Name in Name of option
config/Makefile coqtop -config config/coq_config.ml in configure lib/envars.ml for coqtop/coqdep
--------------------------------------------------------------------------------------------------------
COQLIBINSTALL COQLIB coqlib -libdir coqlib -coqlib
DOCDIR DOCDIR docdir -docdir docdir
CONFIGDIR configdir -configdir
DATADIR datadir -datadir
BINDIR -bindir
MANDIR -mandir
EMACSLIB -emacslib
COQDOCDIR -coqdocdir
Note: in envars.ml, docdir and coqlib are recomputed
|
|
|
|
|
|
|
|
|
|
Universe constraints of the inductive types were not instantiated before being pushed on the environment. This commit fixes this bug.
|
|
|
|
|
|
Deprecations which can't be fixed in 4.02.3 are locally wrapped with
[@@@ocaml.warning "-3"]. The only ones encountered are
- capitalize to capitalize_ascii and variants. Changing to ascii would
break coqdoc -latin1 and maybe other things though.
- external "noalloc" to external [@@noalloc]
|
|
This fixes Théo's bug on eset.
|
|
|
|
|
|
|
|
|
|
|
|
As requested in
https://github.com/coq/coq/pull/384#issuecomment-303809461
|
|
As requested in
https://github.com/coq/coq/pull/384#issuecomment-303809461
|
|
Forwards/backwards reasoning thoughts come from
https://github.com/coq/coq/pull/385#discussion_r111008347
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The ' was originally denoting that we were taking in the projections and
applying the constructor in the conclusion, rather than taking in the
bundled versions and projecting them out (because the projections don't
exist for [ex] and [ex2]). But we don't have versions like this for
[sig] and [sigT] and [sigT2] and [sig2], so we might as well not add the
' to the [ex] and [ex2] versions.
|
|
As per Hugo's request.
|
|
|
|
|
|
As per Hugo's suggestion in
https://github.com/coq/coq/pull/384#issuecomment-264891011
|
|
This tactic does better than [inversion] at sigma types.
|
