From: HENRIKSON, J. <JE...@SA...> - 2005-03-02 19:39:00
|
> Looks like you're using an old version of ExtLib. > There was a temporary fix in the "idup" function of DynArray=20 > which prevent this ocaml bug from happening ( there was a bug=20 > in Obj.dup implementation which must have been fixed on=20 > ocaml-cvs ). Maybe we could make a bug-fix release of extlib=20 > since several patches have been applied ? I downloaded a tarball from sourceforge called extlib-1.3.tgz Is this old? I don't see a newer one posted. Or are you saying I need to download ocaml from CVS? I have their 3.08.1 tarball. Jeff > -----Original Message----- > From: Nicolas Cannasse [mailto:war...@fr...]=20 > Sent: Wednesday, March 02, 2005 11:21 AM > To: HENRIKSON, JEFFREY > Subject: Re: [Ocaml-lib-devel] Extlib.DynArray unstable >=20 >=20 > Looks like you're using an old version of ExtLib. > There was a temporary fix in the "idup" function of DynArray=20 > which prevent this ocaml bug from happening ( there was a bug=20 > in Obj.dup implementation which must have been fixed on=20 > ocaml-cvs ). Maybe we could make a bug-fix release of extlib=20 > since several patches have been applied ? >=20 > Nicolas >=20 > ----- Original Message -----=20 > From: "HENRIKSON, JEFFREY" <JE...@SA...> > To: <oca...@li...> > Sent: Wednesday, March 02, 2005 7:55 PM > Subject: [Ocaml-lib-devel] Extlib.DynArray unstable >=20 >=20 > Hello, >=20 > I find DynArray very unstable on my machine, and will have to=20 > revert my code back to Markus Mottl's Res library for the time being. >=20 > Just calling (of_array [||]) several times will cause a segfault on my > machine: >=20 > Objective Caml version 3.08.1 >=20 > # # #directory "+site-lib/extlib";; > # #load "extLib.cma";; > # open ExtLib;; > # DynArray.of_array [||];; > - : '_a DynArray.t =3D <abstr> > # DynArray.of_array [||];; > - : '_a DynArray.t =3D <abstr> > # DynArray.of_array [||];; > - : '_a DynArray.t =3D <abstr> > # DynArray.of_array [||];; > - : '_a DynArray.t =3D <abstr> > # DynArray.of_array [||];; >=20 > Process caml-toplevel segmentation fault >=20 > It's not completely deterministic. It can require varying=20 > number of calls to get the segfault but usually less than 10.=20 > This test was performed on Linux. I can provide more=20 > details about the configuration if you have trouble=20 > reproducing the problem. >=20 > Regards, >=20 >=20 > Jeff Henrikson >=20 >=20 |