[oe] [PATCH] git: add SRC_URI name

[Bernhard Reutner-Fischer]

On Mon, Feb 22, 2010 at 08:41:59PM +0100, Bernhard Reutner-Fischer wrote:
>On Mon, Feb 22, 2010 at 07:17:54PM +0100, Martin Jansa wrote:
>>On Mon, Feb 22, 2010 at 10:06:00AM -0700, Chris Larson wrote:
>>> On Mon, Feb 22, 2010 at 9:55 AM, Phil Blundell <philb at gnu.org> wrote:
>>> > On Mon, 2010-02-22 at 11:24 -0500, Denys Dmytriyenko wrote:
>>> >> On Sun, Feb 21, 2010 at 04:32:32PM +0100, Bernhard Reutner-Fischer wrote:
>>> >> > Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop at gmail.com>
>>> >> > ---
>>> >> >  recipes/git/git.inc |    2 +-
>>> >> >  1 files changed, 1 insertions(+), 1 deletions(-)
>>> >> >
>>> >> > diff --git a/recipes/git/git.inc b/recipes/git/git.inc
>>> >> > index 644e159..fd7b708 100644
>>> >> > --- a/recipes/git/git.inc
>>> >> > +++ b/recipes/git/git.inc
>>> >> > @@ -2,7 +2,7 @@ DESCRIPTION = "The git revision control system used by the Linux kernel develope
>>> >> >  SECTION = "console/utils"
>>> >> >  LICENSE = "GPL"
>>> >> >
>>> >> > -SRC_URI = "http://www.kernel.org/pub/software/scm/git/git-${PV}.tar.bz2"
>>> >> > +SRC_URI = "http://www.kernel.org/pub/software/scm/git/git-${PV}.tar.bz2;name=git"
>>> >>
>>> >> Is there a patch to add checksums in corresponding recipes to use the name? Or
>>> >> what is the purpose of adding one?
>>> >
>>> > I guess the idea is to make it possible for new recipes (which use
>>> > this .inc file) to include their own checksums.  Seems like a reasonable
>>> > enough plan even if the old recipes are not converted.
>>> 
>>> Has anyone thought about programmatically injecting a name to the
>>> first tarball/zip in the SRC_URI if no sources have that name yet?  It
>>> seems like adding a name to the primary tarball is becoming common
>>> boilerplate, and the common case is for that to be the first source.
>
>For recipes that have only one fetched source, could we perhaps default
>to plain SRC_URI[md5sum], without a name. If a second fetched source is
>added, this would transparently break (i.e. require adding two names).
>>
>>Also would be great to have unified names where possible.
>
>For recipes that have two or more fetched sources, i personally would
>use PN but, as you say, that has the disadvantage that you have to
>potentially lookup the package name in an .inc. Not my call though.
>
>>It would make adding new recipes a bit easier (no need to
>>check/copy&paster name used in .inc and also base.bbclass could easily
>>generate those checksum sections with right name easier.
>>
>>I'm using "archive" as few others also used.
>
>and some use tarball, some pn ....

Not using "archive" as Martin likes as default, but uri%d, but
what do you all think about the attached proposal (including docs!:)?

cheers,

PS: does the ftps fetcher work, i didn't bother to try, i admit..
docs seem to be inconsistent about it WRT wget..
-------------- next part --------------
A non-text attachment was scrubbed...
Name: default_checksums_SRC_URI.00.patch
Type: text/x-diff
Size: 4106 bytes
Desc: not available
URL: <http://lists.openembedded.org/pipermail/openembedded-devel/attachments/20100222/0eaf205a/attachment-0002.bin>