When I run into test failures in www-client/chromium, the normal
procedure is to file a bug and filter the failing tests until the
problem can be fixed (usually upstream).

Ideally, I would like to be able to do the following:

1. Start building chromium using emerge or ebuild.

2. If src_test dies, modify the ebuild to filter the failing tests.

3. Re-run the test phase without rebuilding the source code.

If I do not run ebuild clean, portage re-uses the environment from the
previous run, which does not include my changes to the ebuild.

Is there an easy way to have portage source the modified ebuild
instead of just using the previous environment? Wiping out the
environment file isn't really an option because we set some global
variables in pkg_setup.

On 03/05/2012 08:50 AM, Mike Gilbert wrote:
> When I run into test failures in www-client/chromium, the normal
> procedure is to file a bug and filter the failing tests until the
> problem can be fixed (usually upstream).
>
> Ideally, I would like to be able to do the following:
>
> 1. Start building chromium using emerge or ebuild.
>
> 2. If src_test dies, modify the ebuild to filter the failing tests.
>
> 3. Re-run the test phase without rebuilding the source code.
>
> If I do not run ebuild clean, portage re-uses the environment from the
> previous run, which does not include my changes to the ebuild.
>
> Is there an easy way to have portage source the modified ebuild
> instead of just using the previous environment? Wiping out the
> environment file isn't really an option because we set some global
> variables in pkg_setup.

FEATURES=noauto makes it source the ebuild again, mixing it in with the
previous environment. You can enable it temporarily by exporting the
FEATURES=noauto environment setting to the command:

env FEATURES=noauto ebuild foo.ebuild test
--
Thanks,
Zac

On Mon, Mar 5, 2012 at 12:08 PM, Zac Medico <zmedico@gentoo.org> wrote:
> On 03/05/2012 08:50 AM, Mike Gilbert wrote:
>> When I run into test failures in www-client/chromium, the normal
>> procedure is to file a bug and filter the failing tests until the
>> problem can be fixed (usually upstream).
>>
>> Ideally, I would like to be able to do the following:
>>
>> 1. Start building chromium using emerge or ebuild.
>>
>> 2. If src_test dies, modify the ebuild to filter the failing tests.
>>
>> 3. Re-run the test phase without rebuilding the source code.
>>
>> If I do not run ebuild clean, portage re-uses the environment from the
>> previous run, which does not include my changes to the ebuild.
>>
>> Is there an easy way to have portage source the modified ebuild
>> instead of just using the previous environment? Wiping out the
>> environment file isn't really an option because we set some global
>> variables in pkg_setup.
>
> FEATURES=noauto makes it source the ebuild again, mixing it in with the
> previous environment. You can enable it temporarily by exporting the
> FEATURES=noauto environment setting to the command:
>
> * env FEATURES=noauto ebuild foo.ebuild test
> --
> Thanks,
> Zac
>

That sounds like exactly what I was looking for. Thanks!