devserver: allow calling devserver from anywhere

devserver: allow calling devserver from anywhere

Change-Id: I5b1eaf2d2d2bc1ee51d0ff676e2a64abc98b1322

BUG=5246
TEST=below
1) Modified unittests to run from /
2) Ran unit tests before the change and saw the fail
3) Ran unit tests after the change and saw them succeed
4) sudo FEATURES=test ACCEPT_KEYWORDS=~amd64 emerge cros-devutils (see things installed and unittests passing)

Committed: http://chrome-svn/viewvc/chromeos?view=rev&revision=1347a31

[sosa, 2011-02-08 19:49:47]

your TEST= stanza isn't really sufficient for this change.  Can you start
devserver and try updating from it or run image_to_live from outside the chroot
and make sure it still works?

[sosa, 2011-02-08 19:49:48]

your TEST= stanza isn't really sufficient for this change.  Can you start
devserver and try updating from it or run image_to_live from outside the chroot
and make sure it still works?

[zbehan, 2011-02-08 20:32:23]

Isn't that exactly what devserver_test.py already does? I didn't explore it
very thoroughly but I saw that.

On Tue, Feb 8, 2011 at 8:49 PM, <sosa@chromium.org> wrote:

> your TEST= stanza isn't really sufficient for this change.  Can you start
> devserver and try updating from it or run image_to_live from outside the
> chroot
> and make sure it still works?
>
>
> http://codereview.chromium.org/6425002/
>

[sosa, 2011-02-08 20:36:02]

It's path specific so a script that calls it from a weird location
might be a better test (devserver_test just runs it in place).

On Tue, Feb 8, 2011 at 12:32 PM, Zdenek Behan <zbehan@chromium.org> wrote:
> Isn't that exactly what devserver_test.py already does? I didn't explore it
> very thoroughly but I saw that.
>
> On Tue, Feb 8, 2011 at 8:49 PM, <sosa@chromium.org> wrote:
>>
>> your TEST= stanza isn't really sufficient for this change.  Can you start
>> devserver and try updating from it or run image_to_live from outside the
>> chroot
>> and make sure it still works?
>>
>> http://codereview.chromium.org/6425002/
>
>

[zbehan, 2011-02-10 02:24:36]

Okay, I've modified the test to run in / and call everything using absolute
paths.

I wasn't sure I if i should make a new test that runs in / and keep the old
one as is, but then I thought this will test both cases anyway.

On Tue, Feb 8, 2011 at 9:35 PM, Chris Sosa <sosa@chromium.org> wrote:

> It's path specific so a script that calls it from a weird location
> might be a better test (devserver_test just runs it in place).
>
> On Tue, Feb 8, 2011 at 12:32 PM, Zdenek Behan <zbehan@chromium.org> wrote:
> > Isn't that exactly what devserver_test.py already does? I didn't explore
> it
> > very thoroughly but I saw that.
> >
> > On Tue, Feb 8, 2011 at 8:49 PM, <sosa@chromium.org> wrote:
> >>
> >> your TEST= stanza isn't really sufficient for this change.  Can you
> start
> >> devserver and try updating from it or run image_to_live from outside the
> >> chroot
> >> and make sure it still works?
> >>
> >> http://codereview.chromium.org/6425002/
> >
> >
>

[sosa, 2011-02-10 02:55:23]

LGTM thanks.  Make sure to watch the pfq after you commit this and make sure it
passes.

[zbehan, 2011-02-10 02:57:09]

Does the PFQ depend on devserver in some way? (just wondering)

On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:

> LGTM thanks.  Make sure to watch the pfq after you commit this and make
> sure it
> passes.
>
>
> http://codereview.chromium.org/6425002/
>

[sosa, 2011-02-10 02:59:47]

Yup ... it runs an update test using the devserver.

On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org> wrote:
> Does the PFQ depend on devserver in some way? (just wondering)
>
> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
>>
>> LGTM thanks.  Make sure to watch the pfq after you commit this and make
>> sure it
>> passes.
>>
>> http://codereview.chromium.org/6425002/
>
>

[sosa, 2011-02-10 03:00:51]

INFO: PROGRAM(cbuildbot) -> RunCommand: ['bin/cros_au_test_harness',
'--no_graphics', '--no_delta', '--board=x86-generic',
'--test_prefix=SimpleTest', '--verbose',
'--base_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin',
'--target_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin']
in dir /b/cbuild/src/scripts

This in turn calls cros_run_vm_update -> image_to_live -> starts a dev server.


On Wed, Feb 9, 2011 at 6:59 PM, Chris Sosa <sosa@chromium.org> wrote:
> Yup ... it runs an update test using the devserver.
>
> On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org> wrote:
>> Does the PFQ depend on devserver in some way? (just wondering)
>>
>> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
>>>
>>> LGTM thanks.  Make sure to watch the pfq after you commit this and make
>>> sure it
>>> passes.
>>>
>>> http://codereview.chromium.org/6425002/
>>
>>
>

[zbehan, 2011-02-10 03:02:24]

Awesome.

Unrelated, but I think we should enable FEATURES=test by default, and make
more ebuilds use their test suites.

On Thu, Feb 10, 2011 at 4:00 AM, Chris Sosa <sosa@chromium.org> wrote:

> INFO: PROGRAM(cbuildbot) -> RunCommand: ['bin/cros_au_test_harness',
> '--no_graphics', '--no_delta', '--board=x86-generic',
> '--test_prefix=SimpleTest', '--verbose',
>
>
'--base_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin',
>
>
'--target_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin']
> in dir /b/cbuild/src/scripts
>
> This in turn calls cros_run_vm_update -> image_to_live -> starts a dev
> server.
>
>
> On Wed, Feb 9, 2011 at 6:59 PM, Chris Sosa <sosa@chromium.org> wrote:
> > Yup ... it runs an update test using the devserver.
> >
> > On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org>
> wrote:
> >> Does the PFQ depend on devserver in some way? (just wondering)
> >>
> >> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
> >>>
> >>> LGTM thanks.  Make sure to watch the pfq after you commit this and make
> >>> sure it
> >>> passes.
> >>>
> >>> http://codereview.chromium.org/6425002/
> >>
> >>
> >
>

[zbehan, 2011-02-10 03:03:12]

Awesome.

Unrelated, but I think we should enable FEATURES=test by default, and make
more ebuilds use their test suites.

On Thu, Feb 10, 2011 at 4:00 AM, Chris Sosa <sosa@chromium.org> wrote:

> INFO: PROGRAM(cbuildbot) -> RunCommand: ['bin/cros_au_test_harness',
> '--no_graphics', '--no_delta', '--board=x86-generic',
> '--test_prefix=SimpleTest', '--verbose',
>
>
'--base_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin',
>
>
'--target_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin']
> in dir /b/cbuild/src/scripts
>
> This in turn calls cros_run_vm_update -> image_to_live -> starts a dev
> server.
>
>
> On Wed, Feb 9, 2011 at 6:59 PM, Chris Sosa <sosa@chromium.org> wrote:
> > Yup ... it runs an update test using the devserver.
> >
> > On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org>
> wrote:
> >> Does the PFQ depend on devserver in some way? (just wondering)
> >>
> >> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
> >>>
> >>> LGTM thanks.  Make sure to watch the pfq after you commit this and make
> >>> sure it
> >>> passes.
> >>>
> >>> http://codereview.chromium.org/6425002/
> >>
> >>
> >
>

[sosa, 2011-02-10 03:10:47]

We run with FEATURES=test on all chromeos-base ebuilds using
cros_run_unit_tests already.  We don't set it in build_packages
because many portage packages fail to compile with FEATURES=test and
its recursive.

On Wed, Feb 9, 2011 at 7:02 PM, Zdenek Behan <zbehan@chromium.org> wrote:
> Awesome.
> Unrelated, but I think we should enable FEATURES=test by default, and make
> more ebuilds use their test suites.
>
> On Thu, Feb 10, 2011 at 4:00 AM, Chris Sosa <sosa@chromium.org> wrote:
>>
>> INFO: PROGRAM(cbuildbot) -> RunCommand: ['bin/cros_au_test_harness',
>> '--no_graphics', '--no_delta', '--board=x86-generic',
>> '--test_prefix=SimpleTest', '--verbose',
>>
>>
'--base_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin',
>>
>>
'--target_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin']
>> in dir /b/cbuild/src/scripts
>>
>> This in turn calls cros_run_vm_update -> image_to_live -> starts a dev
>> server.
>>
>>
>> On Wed, Feb 9, 2011 at 6:59 PM, Chris Sosa <sosa@chromium.org> wrote:
>> > Yup ... it runs an update test using the devserver.
>> >
>> > On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org>
>> > wrote:
>> >> Does the PFQ depend on devserver in some way? (just wondering)
>> >>
>> >> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
>> >>>
>> >>> LGTM thanks.  Make sure to watch the pfq after you commit this and
>> >>> make
>> >>> sure it
>> >>> passes.
>> >>>
>> >>> http://codereview.chromium.org/6425002/
>> >>
>> >>
>> >
>
>

[zbehan, 2011-02-10 03:43:00]

PFQ passed, so at least the original case of running from ./ remains
unbroken.

Next stop: Making all scripts run the version from chroot.

On Thu, Feb 10, 2011 at 4:10 AM, Chris Sosa <sosa@chromium.org> wrote:

> We run with FEATURES=test on all chromeos-base ebuilds using
> cros_run_unit_tests already.  We don't set it in build_packages
> because many portage packages fail to compile with FEATURES=test and
> its recursive.
>
> On Wed, Feb 9, 2011 at 7:02 PM, Zdenek Behan <zbehan@chromium.org> wrote:
> > Awesome.
> > Unrelated, but I think we should enable FEATURES=test by default, and
> make
> > more ebuilds use their test suites.
> >
> > On Thu, Feb 10, 2011 at 4:00 AM, Chris Sosa <sosa@chromium.org> wrote:
> >>
> >> INFO: PROGRAM(cbuildbot) -> RunCommand: ['bin/cros_au_test_harness',
> >> '--no_graphics', '--no_delta', '--board=x86-generic',
> >> '--test_prefix=SimpleTest', '--verbose',
> >>
> >>
>
'--base_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin',
> >>
> >>
>
'--target_image=/b/cbuild/src/build/images/x86-generic/latest/chromiumos_test_image.bin']
> >> in dir /b/cbuild/src/scripts
> >>
> >> This in turn calls cros_run_vm_update -> image_to_live -> starts a dev
> >> server.
> >>
> >>
> >> On Wed, Feb 9, 2011 at 6:59 PM, Chris Sosa <sosa@chromium.org> wrote:
> >> > Yup ... it runs an update test using the devserver.
> >> >
> >> > On Wed, Feb 9, 2011 at 6:57 PM, Zdenek Behan <zbehan@chromium.org>
> >> > wrote:
> >> >> Does the PFQ depend on devserver in some way? (just wondering)
> >> >>
> >> >> On Thu, Feb 10, 2011 at 3:55 AM, <sosa@chromium.org> wrote:
> >> >>>
> >> >>> LGTM thanks.  Make sure to watch the pfq after you commit this and
> >> >>> make
> >> >>> sure it
> >> >>> passes.
> >> >>>
> >> >>> http://codereview.chromium.org/6425002/
> >> >>
> >> >>
> >> >
> >
> >
>