Re: I've downloaded oe

[Martin Jansa <martin.jansa@gmail.com>]

On Thu, Nov 25, 2010 at 01:44:13PM +0100, Frans Meulenbroeks wrote:
> 2010/11/25 Klaus Schwarzkopf <schwarzkopf@sensortherm.de>:
> > Am 24.11.2010 10:18, schrieb Frans Meulenbroeks:
> >>
> >> 2010/11/24 Martin Jansa<martin.jansa@gmail.com>:
> >>>
> >>> On Wed, Nov 24, 2010 at 08:52:41AM +0100, Frans Meulenbroeks wrote:
> >>>>
> >>>> What about moving all recipes that do not fetch and also did not fetch
> >>>> during Martins test from last march/april to a directory non-fetching
> >>>> or so?
> >>>
> >>> But those recipes were only those where I wasn't able to find the source
> >>> at all (even with google help and couple of popular premirrors). Any
> >>> origin for source was fine for me, because I was only interested in
> >>> checksum check.
> >>>
> >>>> Apparently these are non fetchable for a long time, with no one
> >>>> apparently building them or having an interest to fix them.
> >>>
> >>> More likely those who were building those before are still building
> >>> OK, because they had it in theirs downloads dir. (but the second part is
> >>> right :/).
> >>>
> >>>> My proposal would be to do that before the release is made, thereby
> >>>> improving the % of the recipes that build.
> >>>
> >>> We can improve it also by putting those missing upstream files to
> >>> sources.openembedded.org and use it as premirror. But first we hav to
> >>> find someone who still have it in downloads dir :/ (I don't).
> >>>
> >>> Removing recipes would need to remove those also from tasks/images and
> >>> in the end release images could miss some important functionaly in worst
> >>> case.
> >>
> >> I think most of the non fetching recipes would fall into one of these
> >> categories:
> >> - recipes that are very specific and are not part of a task or image
> >> - recipes that are older versions and that are not used/build
> >>
> >> Wrt the remark on missing important functionality:
> >> For a user with a clean slate, the functionality is already missing as
> >> the task/image cannot be build from scratch.
> >>
> >> It might be a good idea to do a test trying to build all images for
> >> one distro/machine and see what comes out.
> >> Start with a clean system (empty downloads dir, empty TMPDIR, no
> >> pstaged files etc).
> >> Then bitbake all images. By doing it subsequently without cleaning
> >> TMPDIR inbetween it'll become clear which images do build from scratch
> >> and which don't.
> >>
> >> Is someone able/willing to do this test and report the result (suggest
> >> using distro minimal or angstrom or probably shr and machine angstrom)
> >>
> >> Frans
> >
> >
> > Is it usefull to have an image with all/most recipes? This image can be used
> > for testing the recipes (fetch, compile) for several machines.
> 
> This functionality exists:
> 
> do a
> bitbake world
> or if you only want to fetch:
> bitbake -c fetchall world

Also define
SOURCE_MIRROR_FETCH
to ignore COMPATIBLE_* vars

and still it doesn't catch ie
SRC_URI_machine, SRC_URI_distro

Regards,
-- 
Martin 'JaMa' Jansa     jabber: Martin.Jansa@gmail.com