--FCuugMFkClbJLl1L
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Wed, Mar 31, 2010 at 03:13:32PM +0900, NARUSE, Yui wrote:
> 2010/3/31 NARUSE, Yui <naruse / airemix.jp>:
> > Hi,
> >
> > 2010/3/31 Aaron Patterson <aaron / tenderlovemaking.com>:
> >> On Tue, Mar 30, 2010 at 07:13:32PM +0900, NARUSE, Yui wrote:
> >>> (2010/03/30 17:55), Tanaka Akira wrote:
> >>> I patched rubygems and got following backtrace.
> >>> This is not happen when only Rubygems tests run.
> >>> On test-all, Psych effects rubygems tests.
> >>
> >> I believe this is because both YAML and Psych add Object#to_yaml.
> >>
> >> I am working with Eric to convert Rubygems to take advantage of Psych.
> >> That should clear up the problem.
> >
> > I applied workaround patch in r27125.
>=20
> I noticed this is from API incompatibility:
> YAML.dump( obj, io =3D nil )
> Psych.dump o, options =3D {}
>=20
> def YAML.dump( obj, io =3D nil )
>   obj.to_yaml( io || io2 =3D StringIO.new )
>=20
> def psych_to_yaml options =3D {}
>   Psych.dump self, options
>=20
> def self.dump o, options =3D {}
>   visitor =3D Psych::Visitors::YAMLTree.new options
>=20
> So if Psych.dump can treat io like dump(o, io, options =3D{}),
> it can be fixed fundamentally.
> Is this acceptable solution?

Yes, I think that is fine.  Would you like to apply, or shall I?

--=20
Aaron Patterson
http://tenderlovemaking.com/

--FCuugMFkClbJLl1L
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)

iQEcBAEBAgAGBQJLs6sjAAoJEJUxcLy0/6/Gu5IH/j/u5OLM45FcqMDo8v4ec63R
shCgWSwAGRdqYupj8miWiokzx+Kb8DTcLakN3ocpqPu40zeFPAM/cfwkbr5DIhLh
XogLr8BVCI+kK+vMu2SZBKMddCAJBhH56ORGzm+1wifU9zu8NkprwQ5AkUfeDUP3
DPld3bzDdCchARNlt3hflCbwd/51MoyjG0aRGUNDXFvrcOp8NkXNrLXbxlMPAKw6
xnfhnYDn+9bTxhLXi58/ovD4ZxnjR8+tZW547DeeMQsXfy9eprUxk4hCTjj+GEZX
O9w1BYp7iPBv4xn6IwaPIQzeH79ksiJ0EHk9k7xXRGo5yI2PX/ZLwAk/vgcleQ4=
=Z0yv
-----END PGP SIGNATURE-----

--FCuugMFkClbJLl1L--

On Wed, Mar 31, 2010 at 03:13:32PM +0900, NARUSE, Yui wrote:
> 2010/3/31 NARUSE, Yui <naruse / airemix.jp>:
> > Hi,
> >
> > 2010/3/31 Aaron Patterson <aaron / tenderlovemaking.com>:
> >> On Tue, Mar 30, 2010 at 07:13:32PM +0900, NARUSE, Yui wrote:
> >>> (2010/03/30 17:55), Tanaka Akira wrote:
> >>> I patched rubygems and got following backtrace.
> >>> This is not happen when only Rubygems tests run.
> >>> On test-all, Psych effects rubygems tests.
> >>
> >> I believe this is because both YAML and Psych add Object#to_yaml.
> >>
> >> I am working with Eric to convert Rubygems to take advantage of Psych.
> >> That should clear up the problem.
> >
> > I applied workaround patch in r27125.
>=20
> I noticed this is from API incompatibility:
> YAML.dump( obj, io =3D nil )
> Psych.dump o, options =3D {}
>=20
> def YAML.dump( obj, io =3D nil )
>   obj.to_yaml( io || io2 =3D StringIO.new )
>=20
> def psych_to_yaml options =3D {}
>   Psych.dump self, options
>=20
> def self.dump o, options =3D {}
>   visitor =3D Psych::Visitors::YAMLTree.new options
>=20
> So if Psych.dump can treat io like dump(o, io, options =3D{}),
> it can be fixed fundamentally.
> Is this acceptable solution?

Yes, I think that is fine.  Would you like to apply, or shall I?

--=20
Aaron Patterson
http://tenderlovemaking.com/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.10 (Darwin)

iQEcBAEBAgAGBQJLs6sjAAoJEJUxcLy0/6/Gu5IH/j/u5OLM45FcqMDo8v4ec63R
shCgWSwAGRdqYupj8miWiokzx+Kb8DTcLakN3ocpqPu40zeFPAM/cfwkbr5DIhLh
XogLr8BVCI+kK+vMu2SZBKMddCAJBhH56ORGzm+1wifU9zu8NkprwQ5AkUfeDUP3
DPld3bzDdCchARNlt3hflCbwd/51MoyjG0aRGUNDXFvrcOp8NkXNrLXbxlMPAKw6
xnfhnYDn+9bTxhLXi58/ovD4ZxnjR8+tZW547DeeMQsXfy9eprUxk4hCTjj+GEZX
O9w1BYp7iPBv4xn6IwaPIQzeH79ksiJ0EHk9k7xXRGo5yI2PX/ZLwAk/vgcleQ4=
=Z0yv
-----END PGP SIGNATURE-----