--Nq2Wo0NMKNjxTN9z
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Tue, Jul 03, 2012 at 01:31:53AM +0900, vo.x (Vit Ondruch) wrote:
>=20
> Issue #2565 has been updated by vo.x (Vit Ondruch).
>=20
>=20
> >  AFAIK, the only open issue (that Vit raised [ruby-core:44980]) is that
> >  we shouldn't check in the dummy probe header file.  I agree with Vit, =
so
> >  I want to update my patch before merge. :-)
>=20
> Well, I proposed two options, either have both versions in SCM or keep it=
 out of the SCM. I'm still in favor of the latter, however I realized later=
 that there is one major disadvantage. If we want to avoid the need of Ruby=
 during build from tarball, as it is possible now, then the release manager=
 has to have available either DTrace or SystemTap on his/her system. Not su=
re if that might be fulfilled :/ If not, then both files should be pre-gene=
rated by its maintainer and stored in SCM.
>=20
> In other words, we should think also about associated work-flow and relea=
se management.

I don't think the header file should be generated when the tarball is
being built, but when the user installs Ruby (this is how PostgreSQL
does their dtrace headers).

I'm working on a sed script to convert the .d file to a .h file in the
case that the user doesn't have dtrace on their system.  Then the
release manager doesn't need to have Ruby or DTrace on their system when
making the tar.

How does that sound?

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

--Nq2Wo0NMKNjxTN9z
Content-Type: application/pgp-signature

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (Darwin)

iQEcBAEBAgAGBQJP8g4vAAoJEJUxcLy0/6/GIiIH/3gCUJC2l7k0w8F//OpXIve6
Qb6LBP6adGTYVkEh75ASIzd6OovH5M0BsyUMWFnzKDH6h4tQc5V5ZN9/ZqHtTLKg
i2F36ChyDSkJrVxAXM+no7q8Id2Iu/QPPM0txN4eX/MHnmfPkuq7AHh5TqZPrAoU
qOwS1DVj3j4xk6czQtFDxu8qQTPfZwXSLftimnEKk6fbolgkZLjLiMNKF199Q8JT
ZxErtACkVNwUDeHRl+U3s1RcnLCOnFve4vEnEIfKzO5g2ZCCEZYZgXKF3pJmzX5L
HZhMfTzhGcGyllLzH2hcyxwi6iNRmcjYX/PR79DvhMlzDtVmTOpootduQq5Mh3Y=
=FYuc
-----END PGP SIGNATURE-----

--Nq2Wo0NMKNjxTN9z--

On Tue, Jul 03, 2012 at 01:31:53AM +0900, vo.x (Vit Ondruch) wrote:
>=20
> Issue #2565 has been updated by vo.x (Vit Ondruch).
>=20
>=20
> >  AFAIK, the only open issue (that Vit raised [ruby-core:44980]) is that
> >  we shouldn't check in the dummy probe header file.  I agree with Vit, =
so
> >  I want to update my patch before merge. :-)
>=20
> Well, I proposed two options, either have both versions in SCM or keep it=
 out of the SCM. I'm still in favor of the latter, however I realized later=
 that there is one major disadvantage. If we want to avoid the need of Ruby=
 during build from tarball, as it is possible now, then the release manager=
 has to have available either DTrace or SystemTap on his/her system. Not su=
re if that might be fulfilled :/ If not, then both files should be pre-gene=
rated by its maintainer and stored in SCM.
>=20
> In other words, we should think also about associated work-flow and relea=
se management.

I don't think the header file should be generated when the tarball is
being built, but when the user installs Ruby (this is how PostgreSQL
does their dtrace headers).

I'm working on a sed script to convert the .d file to a .h file in the
case that the user doesn't have dtrace on their system.  Then the
release manager doesn't need to have Ruby or DTrace on their system when
making the tar.

How does that sound?

--=20
Aaron Patterson
http://tenderlovemaking.com/
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.12 (Darwin)

iQEcBAEBAgAGBQJP8g4vAAoJEJUxcLy0/6/GIiIH/3gCUJC2l7k0w8F//OpXIve6
Qb6LBP6adGTYVkEh75ASIzd6OovH5M0BsyUMWFnzKDH6h4tQc5V5ZN9/ZqHtTLKg
i2F36ChyDSkJrVxAXM+no7q8Id2Iu/QPPM0txN4eX/MHnmfPkuq7AHh5TqZPrAoU
qOwS1DVj3j4xk6czQtFDxu8qQTPfZwXSLftimnEKk6fbolgkZLjLiMNKF199Q8JT
ZxErtACkVNwUDeHRl+U3s1RcnLCOnFve4vEnEIfKzO5g2ZCCEZYZgXKF3pJmzX5L
HZhMfTzhGcGyllLzH2hcyxwi6iNRmcjYX/PR79DvhMlzDtVmTOpootduQq5Mh3Y=
=FYuc
-----END PGP SIGNATURE-----