Re: proposing svn write for public repos

--===============0187920673==
Content-Type: multipart/signed; micalg=pgp-sha1;
	protocol="application/pgp-signature"; boundary="GID0FwUMdk1T2AWN"
Content-Disposition: inline


--GID0FwUMdk1T2AWN
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On 02.07.2007 15:24:37, Clifford Wolf wrote:
> Hi,
>=20
> On Mon, Jul 02, 2007 at 01:58:14PM +0200, Tobias Hintze wrote:
> > and personal repositories like:
> >=20
> > rock-linux/trunk/package/blindcoder/
> > rock-linux/trunk/package/clifford/
> > rock-linux/trunk/package/daja77/
> > rock-linux/trunk/package/stefanp/
> > rock-linux/trunk/package/teha/
> > (alphabetical order)
> >=20
> > personal repositories should be writable by the "personal owner", the
> > "public" repo should be writable by a set of developers (blindy, daja,
> > stf, myself)
>=20
> there are 36 "CORE" packages in this repositories - most of them in
> 'public'. I propose that we move them to another repository first..

I suggest 'base'.

> > i know - this allows creation of some preconfig.in files and MIGHT still
> > allow people to do bastardious things... i'm willing to take this risk.
> > (we should restrict write access to people who have proven to be not
> > bastardious in the last years)
> > and if you insist - we can still check for such dangerous files in some
> > svn-hook and disallow the transaction.
>=20
> would you write this hook?

This hook already exists. It is already mentioned in pre-commit.tmpl and
called:
	/usr/share/doc/subversion/tools/hook-scripts/commit-access-control.pl

> The hook should also check for the right commit message format and
> information so there is no conflict with fixauthor.spl (the SPL script
> which is parsing the commit messages and setting the author revision
> property according to it).

IIRC fixautho.spl correctly, that can be done by these two lines in pre-com=
mit:

read line < <( $SVNLOOK log -t "$TXN" "$REPOS" )
[ "${line% *}" =3D=3D "${line}" ] && exit 1

> This is very important! There is no history for revision properties, so
> there also is no way to detect and probably rollback errornous automatic
> updates of this property.
>=20
> > everything else should go through submaster.
>=20
> fine with me since non-CORE patches are simply passed thru atm without re=
al
> testing anyways..


Greetings,
	Benjamin
--=20
Benjamin 'blindCoder' Schieder
Registered Linux User #289529: http://counter.li.org
finger blindcoder-1k+CIuW/[email protected] | gpg --import
--=20
http://www.rocklinux.org/ The Distribution Build Kit

--GID0FwUMdk1T2AWN
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.5 (GNU/Linux)

iD8DBQFGiP1RAsME5O4xgqgRAtCOAJ0UY+6kZVDPQfddVPTeRjLJpd0D6gCgxBYH
Ea724Dd6Ai4NWA8ktKI1lJs=
=WYDp
-----END PGP SIGNATURE-----

--GID0FwUMdk1T2AWN--

--===============0187920673==
Content-Type: text/plain; charset="us-ascii"
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
Content-Disposition: inline

_______________________________________________
rock-devel mailing list
[email protected]
http://www.rocklinux.net/mailman/listinfo/rock-devel

--===============0187920673==--