[openstack-dev] [i18n] [nova] understanding log domain change - https://review.openstack.org/#/c/439500

Sean Dague sean at dague.net
Thu Mar 16 19:18:39 UTC 2017

On 03/16/2017 03:00 PM, Doug Hellmann wrote:
> Excerpts from Doug Hellmann's message of 2017-03-10 09:39:29 -0500:
>> Excerpts from Doug Hellmann's message of 2017-03-10 09:28:52 -0500:
>>> Excerpts from Ian Y. Choi's message of 2017-03-10 01:22:40 +0900:
>>>> Doug Hellmann wrote on 3/9/2017 9:24 PM:
>>>>> Excerpts from Sean McGinnis's message of 2017-03-07 07:17:09 -0600:
>>>>>> On Mon, Mar 06, 2017 at 09:06:18AM -0500, Sean Dague wrote:
>>>>>>> On 03/06/2017 08:43 AM, Andreas Jaeger wrote:
>>>>>>>> On 2017-03-06 14:03, Sean Dague  wrote:
>>>>>>>>> I'm trying to understand the implications of
>>>>>>>>> https://review.openstack.org/#/c/439500. And the comment in the linked
>>>>>>>>> email:
>>>>>>>>> ">> Yes, we decided some time ago to not translate the log files anymore and
>>>>>>>>>>> thus our tools do not handle them anymore - and in general, we remove
>>>>>>>>>>> these kind of files."
>>>>>>>>> Does that mean that all the _LE, _LI, _LW stuff in projects should be
>>>>>>>>> fully removed? Nova currently enforces those things are there -
>>>>>>>>> https://github.com/openstack/nova/blob/e88dd0034b1b135d680dae3494597e295add9cfe/nova/hacking/checks.py#L314-L333
>>>>>>>>> and want to make sure our tools aren't making us do work that the i18n
>>>>>>>>> team is ignoring and throwing away.
>>>>>> So... just looking for a definitive statement on this since there has
>>>>>> been some back and forth discussion.
>>>>>> Is it correct to say - all projects may (should?) now remove all bits in
>>>>>> place for using and enforcing the _Lx() translation markers. Only _()
>>>>>> should be used for user visible error messages.
>>>>>> Sean (smcginnis)
>>>>> The situation is still not quite clear to me, and it would be
>>>>> unfortunate to undo too much of the translation support work because
>>>>> it will be hard to redo it.
>>>>> Is there documentation somewhere describing what the i18n team has
>>>>> committed to trying to translate?
>>>> I18n team describes translation plan and priority in Zanata - 
>>>> translation platform
>>>>   : https://translate.openstack.org/ .
>>>>>   I think I heard that there was a
>>>>> shift in emphasis to "user interfaces", but I'm not sure if that
>>>>> includes error messages in services. Should we remove all use of
>>>>> oslo.i18n from services? Or only when dealing with logs?
>>>> When I18n team decided to removal of log translations in Barcelona last 
>>>> October, there had been no
>>>> discussion on the removal of oslo.i18n translation support for log messages.
>>>> (I have kept track of what I18n team discussed during Barcelona I18n 
>>>> meetup on Etherpad - [1])
>>>> Now I think that the final decision of oslo.i18n log translation support 
>>>> needs more involvement
>>>> with translators considering oslo.i18n translation support, and also 
>>>> more people on community wide including
>>>> project working groups, user committee, and operators as Matt suggested.
>>>> If translating log messages is meaningful to some community members and 
>>>> some translators show interests
>>>> on translating log messages, then I18n team can revert the policy with 
>>>> rolling back of translations.
>>>> Translated strings are still alive in not only previous stable branches, 
>>>> but also in translation memory in Zanata - translation platform.
>>>> I would like to find some ways to discuss this topic with more community 
>>>> wide.
>>> I would suggest that we discuss this at the Forum in Boston, but I think
>>> we need to gather some input before then because if there is a consensus
>>> that log translations are not useful we can allow the code cleanup to
>>> occur and not take up face-to-face time.
>> I've started a thread on the operators mailing list [1].
>> Doug
>> [1] http://lists.openstack.org/pipermail/openstack-operators/2017-March/012887.html
> Based on the feedback on that list, there seems to be no real support
> for maintaining the ability to translate log messages. I therefore
> recommend that teams accept the patches to remove them, and stop
> enforcing their use. Whether or not teams want to make a concerted
> effort to sweep through the code and delete them is up to them.
> Please keep translations for exceptions and other user-facing messages,
> for now.
> As Sean pointed out elsewhere, we can deal with other log-related
> changes independently, since those are likely to need more thought to
> write and review.

ACK. Just landed the stop enforcing patch in Nova -


Sean Dague

More information about the OpenStack-dev mailing list