Subject: Re: CVS commit: pkgsrc/databases/postgresql74
To: Quentin Garnier <cube@cubidou.net>
From: grant beattie <grant@NetBSD.org>
List: pkgsrc-changes
Date: 06/15/2004 07:27:21
--s9fJI615cBHmzTOP
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
Content-Transfer-Encoding: quoted-printable

On Mon, Jun 14, 2004 at 11:18:59PM +0200, Quentin Garnier wrote:

> > forgive me, it's early and I lack caffeine, however, I don't see how
> > this is the cause of the problem described in the PR, or how \s+ will
> > fix it.
>=20
> It clearly is the lack of caffeine.  I fail to see how this can seem
> complicated, but the very purpose of that SUBST block is to replace a line
> in the configure script that affects the localedir variable to an empty
> value by the correct one.  This is explained in the comment just before
> the SUBST block.
>=20
> Problem is, there wasn't enough spaces in the sed expression, so the SUBST
> block did nothing, and localedir stayed empty.  Then locale files get
> installed in /.

I tested this a number of times on Solaris before committing it, if it
wasn't working I wouldn't have done so.

the command used is (was):

	s|\( localedir[ 	:]*=3D\).*|\1${_PKGLOCALEDIR}|

ie, there is no ^ so the fact there is not two spaces but one makes no
difference here. I also don't see how replacing the leading space with
\s+ will make any difference to the outcome of the expression in this
case. (I chose just one leading space to avoid subst'ing it when it's
at the start of a line, that's all.)

note that the 'localedir=3D' line is NOT empty before the sed is done,
it contains "${prefix}/share", so the fact that it is empty afterwards
is a little worrying (${_PKGLOCALEDIR} =3D=3D ""? huh?)

> > did you test this? I didn't think we could use anything more than BREs
> > in SUBST_SED stuff.
>=20
> Yes, I've tested it, however only on NetBSD.  It's hard to keep track of
> what is allowed and what is not depending on your platform of the day, and
> if next time the PostgreSQL run autoconf it adds a space we'll have to
> debug this again.

I don't think that should matter given the above sed expression.

grant.


--s9fJI615cBHmzTOP
Content-Type: application/pgp-signature
Content-Disposition: inline

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.4 (NetBSD)

iD8DBQFAzhg5luYOb9yiFXoRAmfjAJ4vqDYmFAJOiSpa0ObJ5RzoRXnoYwCfXjvO
aI6Mod/eXCXDu37Wlf1GD3U=
=c9+6
-----END PGP SIGNATURE-----

--s9fJI615cBHmzTOP--