← Back to team overview

ubuntu-phone team mailing list archive

  1. ubuntu-phone team
  2. Mailing list archive
  3. Message #15746

Re: Contact sync (call for testing)

  Thread Previous • Date Previous • • Thread Next 
On 22/09/2015 14:56, Oliver Grawert wrote:

hi,

Am Dienstag, den 22.09.2015, 14:07 +0200 schrieb Michael Zanetti:


On 21.09.2015 21:15, Oliver Grawert wrote:

hi,

Am Montag, den 21.09.2015, 15:48 -0300 schrieb Renato Filho:

Hi list,

We are about to land a new architecture for our contact sync. And it
involves a lot of changes on the current service. Because of that we
would like some help from our community to help us testing it before
it lands.


so this is the third time somebody sends a "call for testing" mail to
the public ML without any warning.

can we pretty please make sure there is a gigantic red blinking
disclaimer at the top of such mails that your phone *needs* to be
re-flashed cleanly after such a test *and* that the writability needs to
be reverted manually (a --wipe flash will *not* unset it, you need to
--bootstrap or manually remove the writable flag on disk). this is a
general support list, so we have normal users.
while i agree we need to get some real-life testing from willing people
we really need to make sure they understand the risks, using the citrain
tool or installing silo packages manually breaks any possible upgrade
path, a clean re-flash is unavoidable after such a test run ...


How about adding that warning to the citrain tool instead, or any other
tool that breaks upgrades?


well, half the instructions in this thread don't use the citrain tool
but are manual install instructions for the silo PPA ... phone users
also might not necessarily use ubuntu at all on their host system (and
this not the citrain tool) so in that case it wouldnt help ... it would
really be better to make it clear in the mail (no opposition to add a
warning to citrain additionally indeed).
Agreed. Silo testing can be tricky and even more difficult when additional packages are not pulled as dependencies. Renato, why don't you follow a standard landing process, when you're happy with your testing, think that coverage of automated tests is fine, submit it for QA verification. For such features there is nothing wrong about failing a first landing. Then send a call for testing for broader review once the packages are in rc-proposed. For whoever is willing to help, it will just be a matter of using the right channel but at least the device is not tainted by manual operations. 

JB.


ciao
	oli






--
Jean-Baptiste Lallement
Canonical Services Ltd.
IRC: jibel

References

  Thread Previous • Date Previous • • Thread Next