aboutsummaryrefslogtreecommitdiffhomepage
path: root/library/summary.mli
diff options
context:
space:
mode:
authorGravatar letouzey <letouzey@85f007b7-540e-0410-9357-904b9bb8a0f7>2009-02-03 09:43:13 +0000
committerGravatar letouzey <letouzey@85f007b7-540e-0410-9357-904b9bb8a0f7>2009-02-03 09:43:13 +0000
commitaaf82265aca43a22006e6cf80f1b3cbe1fd594aa (patch)
tree78809c61995d9da37828fa09ac8b063fd158b1c3 /library/summary.mli
parentaf188f7c62ae3cca7620f3738a264c70d2c56597 (diff)
Allow to turn contrib/subtac into a (nat)dynlink'able plugin
Main issue was declare_summary being triggered too late in subtac_obligations, hence the associated init_function was _not_ being done by Lib.init(). Fixed for the moment by an ad-hoc launch of this init_function in subtac_obligations. In other plugins, this issue doesn't appear, since init_function is mostly putting back some empty set into a reference that was initially empty. No need to really run init_function in this case. For future plugins, we will nonetheless have to be careful about that. Of course, the (ref Obj.magic) was not exactly helpful in debugging this matter, see http://caml.inria.fr/mantis/view.php?id=4707 As said by Xavier, naughty naughty boys... git-svn-id: svn+ssh://scm.gforge.inria.fr/svn/coq/trunk@11877 85f007b7-540e-0410-9357-904b9bb8a0f7
Diffstat (limited to 'library/summary.mli')
-rw-r--r--library/summary.mli6
1 files changed, 5 insertions, 1 deletions
diff --git a/library/summary.mli b/library/summary.mli
index 7f300a996..3e6375b0e 100644
--- a/library/summary.mli
+++ b/library/summary.mli
@@ -28,5 +28,9 @@ val section_unfreeze_summaries : frozen -> unit
val module_unfreeze_summaries : frozen -> unit
val init_summaries : unit -> unit
-
+(** Beware: if some code is dynamically loaded via dynlink after the
+ initialization of Coq, the init functions of any summary declared
+ by this code may not be run. It is hence the responsability of
+ plugins to initialize themselves properly.
+*)