| Commit message (Collapse) | Author | Age |
|---|
| | |
|
| | |
|
| | |
|
| |
|
|
|
|
| |
The search algorithm is not satisfactory though, as it crawls through
all known files to find the proper one. We should rather use some trie-based
data structure, but I'll leave this for a future commit.
|
| |
|
|
|
|
| |
Since commit 2f521670fbd, the Require "file" syntax was not used anymore by
coqtop but the code handling it was still there in coqdep. We finish the work
by erasing the remnant code.
|
| | |
|
| |
|
|
| |
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@15715 85f007b7-540e-0410-9357-904b9bb8a0f7
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
There are preliminary problems to solve in the semantics of -I/-R which
differs from the one of coqtop:
- in case of ambiguity, internally to -R option, or between two -R/-I options,
it is not clear at all that the semantics is as in coqtop (which by the way
is not canonical either since it seems to depend on the underlying order of
dirs in the file system),
- in the presence of -I dir, Require A.b does not look if dir.A.b exists;
similarly for Declare ML Module "f", Require "f" and Load "f" when f has
a dirname part.
I suggest to have a common library for coqdep, coqdoc and coqtop that
ensures that -I/-R behave consistently for all tools.
Incidentally made coqdep lexer a bit more strict about syntax errors.
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14626 85f007b7-540e-0410-9357-904b9bb8a0f7
|
|
|
Sequel to commit 14476 : in fact, even with "tools" in .PHONY,
we still may have coqdep stuff being recompiled in a "make"
that follows a successful "make". This seems to related to
the hacks I've introduced to prevent ocamlopt from erasing
and recreating .cmi when there's no .mli around
(cf. comment around line 823 in Makefile.build).
Scenario:
- First we build coqdep_boot directly out of coqdep_lexer.ml and co.
When ocamlopt is around, this creates some .cmx and .cmi,
but no .cmo.
- Later we build coqdep, which need coqdep_lexer.cmx and co.
Now "make" checks whether these .cmx are up-to-date.
But our hacks made these .cmx depend on the corresponding .cmi.
Then "make" checks whether these .cmi are up-to-date.
But our hacks made these .cmi depend on the corresponding .cmo.
These .cmo doesn't exist yet, we run ocamlc, which recreates
the .cmi with same content but a different timestamp.
For some strange reason, even with refreshed .cmi, the .cmx
are not remade by this run, but will be on the next run.
Conclusion: what a mess. Implicit rules about .cmx / .cmo / .cmi
should be improved, but I see currently no simple solution.
In the meantime, an simple ad-hoc fix is to create these two .mli ...
git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@14479 85f007b7-540e-0410-9357-904b9bb8a0f7
|
Théo Zimmermann
Pierre-Marie Pédrot
Maxime Dénès
herbelin
letouzey